Escolar Documentos
Profissional Documentos
Cultura Documentos
Part I
c
2008
Herman Mann
Czech Technical University in Prague
herman.j.mann@gmail.com
Contents
1
2 CONTENTS
All aspects of modern technology are underlined by the subject of system dynamics, which plays the
determining role in the market competition of engineering products. Its importance increases with the
ever-growing demands on operational speed, energetic efficiency, safety, reliability, and environment pro-
tection. At the same time, the engineering products become more and more complex with regards both
to the number of their components as well as to the variety of phenomena affecting the product dynamics,
either in a useful or undesirable way.
To understand and predict dynamic behavior of existing systems as well as to design new systems,
engineers resort to computer-assisted modeling, simulation and analysis of the systems. Modeling is
in this context a procedure for forming a simplified representation of those features of real dynamic
systems that are relevant to their dynamic behavior. The simplification is based on an abstraction of
the relevant system features and on their idealization. The resulting model representation can take the
form of equations, diagrams, or tables of data, for example. Simulation is a procedure imitating dynamic
behavior of the modelled real systems. Analysis is a procedure investigating changes of the system
behavior if the system or its surroundings are changing so it gives more thorough information about the
system dynamics than simulation.
Such techniques can replace to a certain degree expensive and time consuming design verification
based on construction of system prototypes and their experimental investigation using measuring instru-
ments. At the same time, these techniques allow engineers considering a larger variety of system design
alternatives and their more thorough verification. Design verification by simulation is the only option in
cases where the experimentation with real systems would be too expensive or too dangerous. Analysis
can also simplify the system operation and maintenance by detecting an impending system failure as well
as by helping to locate a fault in the system and to diagnose its cause. As these computation techniques
allow for faster introduction of a new product into the market as well as for attaining and maintaining
the product higher quality, they help in acquiring higher profits.
This Web-based course on dynamics of multidisciplinary controlled systems has been developed for
• industrial enterprisers interested in enhancing the qualification and efficiency of their personnel
The subject of dynamics and control underlies all aspects of modern technology and plays the de-
termining role in the world-market competition of engineering products. Its importance increases with
the ever-growing demands on operational speed, efficiency, safety, reliability, or environment protection.
Nevertheless, entrepreneurs report lack of professionals sufficiently qualified in this field and vocational
schools complain about the overall decline of interest in engineering study among young people. Profes-
sional associations call for radical changes in the engineering curriculum and for innovative approaches
to vocational training.
The existing courses on dynamics and control are criticized for discouraging young people from en-
gineering study by overemphasizing theory and mathematics at the expense of practical engineering
issues. Dynamics is usually covered in several courses separated along the borders between the tradi-
tional engineering disciplines despite the fact that most of the contemporary engineering products are of
3
4 CONTENTS
multidisciplinary nature. Control courses are criticized for presenting ’textbook’ problems engineered to
fit the theory without undertaking a realistic modeling of the controlled systems. Computers are often
used to carry out old exercises without radical modification of the curriculum to fully incorporate current
computer capabilities.
The contents, emphasis and style of this course differs from most of the existing courses by a number of
factors. The first part of the course exposes learners in a systematic and unified way to realistic modelling
and simulation of multidisciplinary systems. This approach is supported by a free online access to the
simulation software DYNAST. It allows for submitting dynamic models directly in a graphical form
isomorphic to the geometric configuration of the modelled real systems. The underlying equations are
not only solved, but also automatically formulated.
Using this tool, learners can be introduced to dynamics through modeling and simulation of simple
yet practical examples. This stimulates their interest before they are exposed to more abstract subjects
like equation formulation, block-diagram construction, Laplace transformation, etc. In addition, learners
are given in the course a better ‘feel’ for system dynamics by graphical visualization, 3D animation and
interactive virtual experiments. Computers are thus integrated into the course curriculum from the very
start as learning tools without distracting learners from dynamics by issues not related directly to it.
In the same way, learners are given the hands-on opportunity to acquire the skills necessary for efficient
solving real-life problems in their engineering profession.
Different target groups can select individual paths through the course, paths tailor-made to their
previous experience and respecting their learning needs. Depending on their background, learners can
choose to start from the module on dynamics of either mechanical, electrical or fluid systems. More
advanced learners can go directly to the dynamics of multidisciplinary systems, or even to control. The
course material as well as the learning environment suit both to self-study and remote tutoring. They
both strongly support investigative and collaborative modes of learning.
Learners are given the opportunity to benefit from ‘organizational learning’, i.e. from utilizing know-
ledge recorded during previous problem solving both in academia and industry. This knowledge is avail-
able to learners in the form of a large collection of online re-solvable examples and submodels of systems
components and physical phenomena.
References
Multidisciplinary systems
R. H. Cannon: Dynamics of Physical Systems. McGraw-Hill, New York 1967
D.C. Karnopp, D.L. Margolis, R.C. Rosenberg: System Dynamics: A Unified Approach. John Wiley,
New York 1990 (2nd ed.)
M.M. Tiller: Introduction to Physical Modeling with Modelica. Kluwer Academic Publishers 2001
P. Gawthrop, L. Smith: Metamodelling: Bong graphs and dynamic systems. Prentice Hall 1996.
F.T. Forbes: Modeling and Analysis of Engineering Systems, a Unified Graph-Centered Approach,
Marcel Dekker 2001.
J.L. Shearer, B.T. Kulakowski, J.F. Gardner: Dynamic Modeling and Control of Engineering Systems.
Prentice Hall, 1997.
R.L. Woods, K.L. Lawrence: Modeling and Simulation of Dynamic Systems. Prentice Hall, 1997.
I. Cochin, W. Caldwallender: Analysis and Design of Dynamic Systems. Addison-Wesley, 1997 (3rd.
ed.)
W.J. Palm III: Modeling, Analysis, and Control of Dynamic Systems. John Wiley & Sons, 1999.
Fano, R.M., Chu, L.J.: Fields, Energy, and Forces. Wiley, New York 1959.
M.H. Rashid: SPICE for Circuits and Electronics Using PSpice. Prentice-Hall Int. Editions, Engl. Ciffs
1990.
Slemon G.R.: Magnetoelectric Devices. Transducers, transformers, and machines. John Wiley, New
York 1966.
R. Kielkowski: Inside SPICE. Overcoming the Obstacles of Circuit Simulation. McGraw-Hill, New York
1994.
J.A. Connelly, P. Choi: Macromodeling with SPICE. Prentice Hall, Engl. Cliffs 1992.
M.H. Rashid: Power Electronics. Circuits, Devices, and Applications. Prentice-Hall Int. Editions, Engl.
Ciffs 1988.
J. Vlach, K. Singhal: Computer Methods for Circuit Analysis and Design. Van Nostrand Reinhold 1994
(2nd ed.).
N. Mohan, T.M. Undeland, W.P. Robbins: Power electronics. Converters, Applications, and Design.
Wiley 1995 (2nd ed.).
Mechanical systems
P.E. Nikravesh: Computer-Aided Analysis of Mechanical Systems. Prentice-Hall 1988
E.J. Haug: Computer Aided Kinematics and Dynamics of Mechanical Systems. Allyn and Bacon 1989.
J.L. Meriam, L.G. Kraige: Engineering Mechanics. Statics. Dynamics. Wiley 1998 (4th ed.).
F.S. Tse, I.E. Morse, R.T. Hinkle: Mechanical Vibrations. Theory and Applications. Prentice-Hall
1978.
J.J. Craig: Introduction to Robotics. Mechanics and Control. Addison-Wesley 1989 (2nd ed.).
DUBBEL Handbook of Mechanical Engineering. Springer 1994.
H. Mayr: Virtual Automation Environments. Design, Modeling, Visualization, Simulation. Marcel
Dekker 2002.
Electro-mechanical systems
A. Lenk, G. Pfeifer, R. Wertschtzky: Elektromechanische Systeme. Mechanische und akustische Net-
zwerke, deren Wechselwirkungen und Anwendungen. Springer 2001.
W. Leonhard: Control of Electrical Drives. Springer 1985.
P. Krause, D. Wasynczuk, S. D. Sudhoff: Analysis of Electric Machinery. IEEE Press 1995.
H. Gross: Electrical Feed Drives for Machine Tools. Wiley 1983.
P.C. Krause: Analysis of Electric Machinery. McGraw-Hill 1987.
H.H. Woodson, J.R. Melcher: Electromechanical Dynamics. Part I: Discrete Systems. Part II: Contin-
uum Electromechanics. Part III: Elastic and Fluid Media. Wiley 1968 [repub. by Krieger 1990].
J. Meisel: Principles of Electromechanical-Energy Conversion. McGraw-Hill 1966.
D.C. White, H.H. Woodson: Electromechanical Energy Conversion. Wiley 1959.
T. Kenjo: Stepping motors and their microprocessor controls. Clarendon Press 1992.
R.T. Smith: Analysis of Electrical Machines. Pergamon Press 1982.
S.A. Nasar, L.E. Unnewehr: Electromechanics and Electric Machines. Wiley 1979.
J.R. Melcher: Continuum Electromechanics. MIT Press 1981
J.H. Lienhard: A Heat Transfer Handbook. Prentice-Hall, Inc., Englewood Cliffs, N.J., 1981.
Y.A. Çengel: Thermodynamics. An Engineering Approach. McGraw-Hill, New York 1994 (2nd ed.).
H.E. Meritt: Hydraulic Control Systems. Wiley 1967.
J. Watton: Fluid Power Systems. Modelling, simulation, analog and microcomputer control. Prentice
Hall 1989.
Module 1
Introduction to dynamics
Module units
1.1 Modeling and simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-1
1.2 Role of simulation in engineering . . . . . . . . . . . . . . . . . . . . . . . . . 1-2
1.2.1 Design problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-2
1.2.2 Design process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-4
1.2.3 Design descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-4
1.2.4 Design levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-6
1.2.5 Design methodologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-7
Module overview. Welcome to the course part focused on system dynamics. It is well recognized
that computer-assisted modeling and simulation are indispensable for understanding dynamics of existing
systems as well as for designing new systems. This module attempts to explain why is it so.
1-1
1-2 MODULE 1. INTRODUCTION TO DYNAMICS
We shall see that there are many striking similarities between mechanical, electrical, magnetic, fluid,
acoustic, and thermal systems, which will permit a unified study of the system behavior. A uniform
terminology and symbolism applicable to all of the physical systems considered will be developed. The
integration of subject matter in this course is designed to allow the learners to recognize that a large
class of problems can be considered in a similar and systematic fashion.
In the next step, the engineers must define the system to be considered (should the ambient-temperature
variation be considered an input? is the inertia of a connecting shaft important in this situation? etc.).
They then may describe the system by means of the various physical elements or more complex submodels
which we will consider. After this, they must investigate the energetic interactions between these system
parts.
The next job is to establish a mathematical model of the system to be analyzed. This involves the
identification and idealization of individual system components as well as identification and idealization
of their interconnection. The mathematical statement of the governing relationships between system
variables is called the formulation problem. Interconnection of the elements imposes constraints on the
variation of system variables, and the convenient way of specifying these constraints is by a mathematical
statement of the way in which the various variables are related.
To investigate the dynamic behavior of the system, this set of equations must be solved. The initial
state of the system must also be specified by a set of initial conditions. The system equations must be
solved for the desired response under the specified input signal and initial conditions. There are numerous
approaches to the solution of the system equation, and the choice of a method or software will depend
on the problem at hand.
After obtaining a solution for the response, the engineers have another major job. The tasks of initial
modeling and final analyzing or designing are functions which clearly distinguish the engineer from the
mathematician. The engineers must check their solution (is it correct dimensionally? does it correspond
to physical reality? does it check for simplified situations which can be easily analyzed? etc.). Their
interest in the whole matter of dynamic systems is to eventually analyze to determine whether a certain
performance is obtained and is satisfactory, or more generally, he must design the system so that it will
meet certain performance specifications.
In summary, our approach to a dynamic system problem can be enumerated as follows:
6. analyze or design.
?
MANUFACTURING
?
OUTPUT TESTING
?
SYSTEM OPERATION
?
MAINTENANCE
?
SYSTEM OPERATION
tolerances. There are two basic approaches to the problem: worst-case and statistical tolerance design.
Worst-case design aims at preventing system failure even in the case of the most adverse combinations
of the parameter deviations. Very few systems, however, are designed for 100% production yield, i.e.,
that all 100% ‘copies’ of the system are correct at the output of the production process. The reasons
are economic one: it is cheaper to dump or repair a part of the production than to utilize components
with tolerances more narrow. The aim of statistical tolerance design is to achieve the required system
production yield with components of given statistical distribution of their tolerances.
Of course, a system may fail also if one or more of the system components fail, i.e. if the properties
of these components get outside their admissible tolerances. The probability that the system would
not fail in this way during certain time period and under certain operating conditions is called system
reliability. The objective of reliability design is to meet the requirements on the system reliability taking
into consideration the limited reliability of the system components.
Design for testing aims at designing systems in such a way that they could be tested easily, some
systems are even designed to be self-testing. The design for testing is complemented by design of tests.
As Fig. 1.1 suggests, different tests are needed for design verification, for testing at the output of the pro-
duction, or for testing of systems during their in-service operation and maintenance. Such experimental
testing is increasingly carried out using computer-controlled instruments.
The objectives of system testing can be of different levels of depth:
• fault localization reveals in which of the system components the failure occurred
There are also other design activities, like design for producibility, for maintainability, for reproducibil-
ity, for built-in obsolescence, for self-destruction, etc.
In all these activities computer-assisted modeling, simulation and analysis makes an indispensable
tool today. Used to explore and predict dynamic behavior of dynamic systems, it plays a key role in
enhancing their quality and reliability. Simulation also decreases the need for costly and time consuming
1-4 MODULE 1. INTRODUCTION TO DYNAMICS
DESIGN DESCRIPTIONS:
behavioral topological geometrical
DESIGN LEVELS:
conceptual
functional
DESIGN SPACE
physical
technological
production of real prototypes and for their experimental testing. Many of these activities cannot be
performed experimentally at all, like the tolerance or reliability design, for example. Last but not least,
by speeding up the design process, simulation allows for acquiring higher profits by introducing new
products into the market faster.
(a)
C D
x
(b)
. D
A S
C
B
W N
f
(c)
Figure 1.3: (a) Copying lathe, (b) lathe decomposition, (c) multipole model.
1-6 MODULE 1. INTRODUCTION TO DYNAMICS
The behavioral description characterizes external dynamic interactions of a module in terms of the rela-
tionship between the interactions ignoring completely the module internal structure, e.g., it characterizes
the module just as a ‘black-box’. The model, which the behavioral description of a module represents,
may be static or dynamic, linear or nonlinear, time-variable or time invariant, lumped-parameter or
distributed-parameter, etc. The description may be in the form of various characteristics obtained by
experimenting with the real module or in the form equations, transfer functions, state-transition tables,
etc., based on a hypothesis about the physical effects taking place within the module. The experimental
and hypothetical approaches are usually combined.
On the other hand, to represent mutual interactions between modules in a system, or between submod-
ules in a module, is the aim of the topological description. Any module geometrical shapes, dimensions
and positions are meaningless in this description, only the coincidences of the module interactions are
relevant. The topological descriptions are usually portrayed graphically in the form of diagrams in which
modules or their functions are represented by some symbols and the module interactions by line segments
interconnecting the symbols.
Fig. 1.3c gives an example of a topological description of the copying lathe. Besides the standardized
symbols for a spool valve and a hydraulic cylinder, we can see there a damper bf modeling the friction
between the cylinder and the slide, the combination of damper bW and spring kN modeling the dynamics
of the manufacturing process, the actuator of velocity ṙ enforced to the stylus by the rotating master,
the square symbol representing inertial mass m of the movable parts, and the symbol S representing
there for the oil pressure supply. The interactions between the modules are denoted in Fig. 1.3c by
small circles. Thus the letters A and B denote interactions carried out by mechanical interlinking of the
modules whereas C and D denote interactions between the module fluid inlets.
• Each of the line segments interconnecting blocks in a block diagram represents an unidirectional
transfer of one mathematical variable. The line segments interconnecting multipoles in a multipole
1.2. ROLE OF SIMULATION IN ENGINEERING 1-7
Figure 1.4: Copying lathe block diagrams at (a) the conceptual and (b) functional design level.
diagram represent bilateral transfer of energy between system modules via electrical wires, fluid
lines, shafts or other mechanical links, etc. Each of the line segments is associated with two physical
quantities the product of which is a physical dimension corresponding to power.
• The mutual interactions of blocks follow algebraic rules whereas the multipole interactions are
governed by physical laws of energetic balance and geometric compatibility.
• The block diagram is just a graphical representation of equations. To set up this diagram these
equations must be formulated first and then converted into the diagram. The structure of this
diagram has nothing in common with the structure of the real system which the equations charac-
terize. The topological structure of the multipole diagram corresponds directly to the geometrical
structure of the real system. Thus setting up of this diagram can be based simply on inspection of
the real system.
At the technological design level the assembly or another manufacturing process for the system and
for its components is designed.
Bodies of the cars constitute together with the body of the rod a mechanical rectilinear system.
Behavior of the system is influenced by the system surroundings formed by the wind blowing opposite to
the car motions as well as by the friction between the car tyres and the road surface.
Fig. 2.2 was sketched for the purpose of the dynamic analysis focused on the behavior of a car-wheel
suspension. When the car travels along a road, the wheel tyre copies potholes in the road surface. The
wheel is thus driven up or down in the vertical direction along the axis y. In this case, the rectilinear
system under consideration consists of four bodies: the spring, the shock absorber, the wheel and the
quarter of the car body. The system excitation by the road surface and the gravitational attraction of
the wheel and the quarter-car body forms the system surroundings in this case.
2-1
2-2 MODULE 2. RECTILINEAR MECHANICAL SYSTEMS
If a body can be considered rigid its rectilinear motion is completely specified by the motion of any
one point in the body, since all points have the same motion. A rigid body is a body whose changes in
shape are negligible compared with the overall dimensions of the body or with the changes in position of
the body as a whole. If each of the cars shown in Fig. 2.1 is considered as a rigid body, motions of the
cars can be represented, for example, by the points A and B denoting the cars’ rod hooks.
Absolute rectilinear motion of a given point is defined as the motion of that point relative to
a fixed point in space considered as a reference frame for the motion. Because of the difficulty of
providing a real fixed point in space from which to make measurements, the concept of absolute motion
is strictly hypothetical, but nevertheless very useful.
A reference frame fixed to the earth is often a reasonable approximation. However, when dealing with
the motion of long-range missiles, for example, the earth cannot be considered to be a nonaccelerating
reference and velocities are then measured with respect to the ‘fixed’ stars instead.
Fig. 2.4 shows a geometric configuration of points A and B representing the car hooks in Fig. 2.1.
Motion of the points is constrained to the rectilinear translation along the x-axis ‘painted’ on the reference
frame. The position xA (t) of point A is defined as the distance of this point from the origin 0 chosen
on the axis x. In the figure, xA (t) is shown negative whereas the position xB (t) of point B is positive.
Positions are measured in meters [m].
Figure 2.4: (a) Positions and (b) velocities of points in rectilinear motion.
The absolute velocity ẋA (t) [m/s] and the absolute acceleration ẍA (t) [m/s2 ] of point A is by
definition
d dẋA d2 xA
ẋA (t) = xA and ẍA (t) = =
dt dt dt2
We shall depict the absolute velocities in drawings of geometric configurations by empty-head arrows
as shown in Fig. 2.4. Displacements, velocities and accelerations are taken as positive if they correspond
to a motion in the chosen positive direction of the related coordinate axis.
We shall often be concerned with the relative motion of two points. The relative displacement xBA ,
relative velocity ẋBA and relative acceleration ẍBA of point B with respect to point A is at any instant
xBA = xB − xA , ẋBA = ẋB − ẋA , and ẍBA = ẍB − ẍA (2.1)
Position, velocity or acceleration of a point with respect to another point can be measured by a device
applied to the two points “across” the system without braking into it. Velocity can be thus considered as
2.2. RECTILINEAR INTERACTIONS 2-3
an across variable, position as an integrated across variable, and acceleration as a differentiated across
variable.
2.2.2 Forces
Force is the action on a body characterized by its magnitude, by the direction of its action, and by its
point of application. Forces are measured in physical units called newton [N]. A force tends to move a
body in the direction of its action.
Forces acting on a body are classified as either contact forces or body forces. Contact forces are
generated through direct physical contact between two bodies or between a body and a fluid. Body
forces are those applied by remote action, such as gravitational, electrostatic or electromagnetic forces.
Actually both contact and body forces are distributed forces as the contact forces are in fact distributed
over the cross-section area of the contact. In the system shown in Fig. 2.1, the forces of the wind acting
on the cars are distributed over the car surfaces, while the forces in the towing rod are distributed over
the rod cross-section area. If a rigid body is constrained to rectilinear translation all the forces acting on
the body can be considered as concentrated forces acting at a point.
To take a complete account of all forces acting on a body, it is essential to construct the free-body
diagram by isolating the body completely from the rest of the system and its surroundings. This diagram
consists of a closed outline of the external boundary of the body. The diagram shows all the external
contact and body forces applied to the body.
Fig. 2.5a shows the free-body diagram of a towed car, a part of the rectilinear system introduced in
Fig. 2.1. Ft is the force exerted on the car by the towing rod, Fw represents the distributed force of wind
acting on the car surface, and Ff is the friction force acting on the car wheels due to the roughness of
the road.
As it is indicated by full-head arrows, we have chosen the following convention for the orientation of
forces applied to a body. The assumed positive direction of all applied forces is in agreement with the
direction of the reference axis. Values of those applied forces that tend to accelerate the body in this
direction are positive, if an applied force tends to decelerate the body its value is negative.
When we are dealing with an external action of a force on a rigid body in rectilinear motion, the force
may be applied at any point of the body without changing its effect on the body as a whole. Thus, if the
car shown in Fig. 2.5a is considered as a rigid body, all three forces applied on it can be replaced by one
concentrated force Fa acting for example at point A as shown in Fig. 2.5b. Obviously,
Fa = Ft + Fw + Ff (2.2)
A force F can be viewed as the time rate of the linear momentum µ [N.s] (called also impulse)
acting in the same direction, i.e.
d
F (t) = µ = µ̇ (2.3)
dt
or
Z t
µ= F (t) dt + µ0 (2.4)
0
where ta and tb are the beginning and end of the time interval over which we desire to compute the work
done.
The mechanical energy E supplied to a body is the sum of the work done on the body by all forces
acting on it
Xn
E= (Wab )k (2.8)
k=1
These relations represent geometric constraints which are placed on the collinear rectilinear mo-
tions of points due to geometric continuity of space. This is just a demonstration of the postulate of
compatibility in rectilinear translation.
Re-solvable example
Composition of motions
This shows that the forces F1 (t) and F2 (t) are of the same magnitude but of the opposite direction at
any instant. If F1 is considered as an action force, F2 then forms the reaction force with respect to F1 ,
and vice versa.
The action of a force on a body can be separated into two effects, external and internal. Forces
external to a body are of two kinds, applied forces and reactive forces. The effects of applied forces
internal to the body are the resulting internal stresses and strains distributed throughout the material of
2.3. COMPOSITION OF RECTILINEAR INTERACTIONS 2-5
Figure 2.6: Balance of forces acting on (a) a point, (b) on a body, (c) on an ideal link, (d) on a configuration
of ideal links.
the body. The relation between internal forces and internal strains involves the material properties
of the body and is studied in strength of materials, elasticity, and plasticity. In mechanics of rigid bodies,
concern is given only to the net external effects of forces. Fig. 2.6b shows the external force Fa applied
on a car and the corresponding reaction force of the car Fc = Fa .
The principle of transmissibility states that an external force applied at a point of a rigid body
constrained to rectilinear motion may be considered to be transmitted or to flow unchanged through the
body to any other point of the body.
Fig. 2.6c shows points A and B moving along the x-axis. The points, which are interconnected by an
ideal link, are acted on in the x-direction by forces F1 and F2 external to the link. The ideal rectilinear
link is by definition absolutely rigid and has no mass. The link is characterized by the relations
F1 (t) + F2 (t) = 0 (2.11)
ẋA (t) = ẋB (t) (2.12)
The relation (2.12) is identical with (2.10). This indicates that due to the principle of transmissibility
the force equilibrium holds also if the forces F1 (t) and F2 (t) applied to the endpoints of an ideal rectilinear
link. The second relation (2.12) follows from the assumption that there are no link deformations due to
forces acting on the link. As the link is massless it is not subjected to any inertial forces and it neither
accumulates, nor dissipates any energy.
Figure 2.7: (a) Applied and reaction forces. (b) Reaction forces flowing through the system. (c) Reaction
forces in the pair of scales.
This example may be generalized for any number of points interconnected by ideal links. Fig. 2.6c
shows three points A, B and C acted on by forces F1 , F2 and F3 , respectively. The points are linked to
2-6 MODULE 2. RECTILINEAR MECHANICAL SYSTEMS
point D. All the points are in rectilinear translation along the x-axis. The three forces are in equilibrium
and the point velocities are identical, i.e.,
Fig. 2.7a shows two cars interconnected by a towing rod. The cars are considered as rigid bodies and
the rod as an ideal rectilinear link. The assume positive orientation of the car reaction forces Fc1 and Fc2
shown within the car outlines opposes the x-direction. The car reaction forces, which can be imagined
as flowing through the two-car assembly, are in balance, i.e.
To prove (2.15), let us consider Fig. 2.7b showing free-body diagrams of the cars and towing rod. The
forces denoted outside the cars and the rod are applied forces the assumed positive direction of which is
the x-direction. Let us assume that
where Fr1 and Fr2 are forces applied on the cars by the towing rod, and Fo1 and Fo2 are the other external
forces applied on the cars. When submitting these relations into (2.15) after taking into consideration
that Fc1 = Fa1 and Fc2 = Fa2 we obtain
The forces Fr1 and Fr2 acting on the rod are in balance, and the other external forces Fo1 and Fo2 applied
to the system of rigid bodies must be in balance too.
The reaction forces in the vertical links of the pair of scales shown in Fig. 2.7c must be balanced, i.e.
Fa + Fb + Fc = 0
Re-solvable example
Static equilibrium of forces
where µ is the linear momentum acting on the mass. By definition, µ = 0 when v = 0. Thus
Z t
mv = F (t) dt + µ(t0 ) (2.18)
t0
The inertia effects encountered in rectilinear translation of bodies can be modelled by the idealized
physical element called pure inertor. The inertor ignores, however, damping, spring or any other
2.4. RECTILINEAR INERTIA OF MASS 2-7
Figure 2.8: (a) Symbol of inertor. (b) Momentum vs. velocity characteristic of ideal inertor. (c) Sliding
body, (d) its model.
energy effects in bodies. Our symbol for the inertor is shown in Fig. 2.8a. The symbol consists of a
square representing the mass and of a short line segment called a pin. The pin denotes the pole of the
inertor which represents the inertor energy interactions with the rest of the system.
The pure inertor is associated with two variables: the inertor force F representing the inertia force
of the modelled mass, and the inertor velocity v representing the absolute velocity of the mass. The
assumed positive orientation of F and v with respect to the inertor symbol is indicated in Fig. 2.8a by
the full-head and empty-head arrow, respectively.
Dynamic behavior of the pure inertor is completely described by the constitutive relation
µ = f (v) (2.19)
where f (·) is a function. A pure inertor characterized by the linear constitutive relation (2.17) or (2.16),
where the mass m is constant, is called the ideal inertor. The relationship (2.17) is plotted in Fig. 2.8b.
From the theory of relativity we know that mass of a given body increases with the body velocity
approaching the velocity of light. As the present considerations are limited to velocity levels small
compared with the velocity of light, Newton’s second law can be applied. No other nonlinear inertia
effects are known. In some engineering problems, however, the mass of a body changes with time.
Fig. 2.8c shows the free-body diagram of a rigid body sliding with a negligible friction along a hori-
zontal surface in the x-direction. Let us assume that point A was chosen to represent the body rectilinear
motion as well as the site of application of all the external forces Fext acting on the body. Such a point
can be considered as the body mass point, i.e. a point of the body associated with the lumped body
mass constrained to the rectilinear motion.
Fig. 2.8d gives the dynamic model of the sliding body in the form of a multipole diagram. Point A of
the real body moving in the x-direction as shown in Fig. 2.8c is represented in the diagram by the node
Ax . Diagram nodes represent sites of energy interaction between bodies.
This is in agreement with our convention about the orientation arrows for reactions to external forces
applied to bodies. The empty-head arrow, which denotes the assumed positive orientation of the inertor
velocity v, is directed towards the inertor pole away from the reference frame.
As we assume that the body is not subjected to any other effects except its own inertia, there is only
one physical element in the diagram – the inertor – which is attached to node Ax . If no external force
Fext is acting on the body in this example, the inertia force Fi is zero and the body velocity ẋA is also
zero or constant.
The inertor is in fact a two-pole physical element. But its second pole, with respect to which the
inertor velocity should be ‘measured’, must be always attached to the reference frame. For this reason,
the second pole need not be shown in the inertor symbol just to simplify the diagrams.
2-8 MODULE 2. RECTILINEAR MECHANICAL SYSTEMS
If P > 0, energy is being stored in the inertor, and if P < 0, the stored energy is being retrieved from
it. Which of these two states is occurring depends both on the relative velocity and acceleration at the
moment.
Integration of (2.20) gives the expression for the energy accumulated in the ideal inertor
1
E= m v2
2
which is called rectilinear kinetic energy. The energy is represented by the area in Fig. 2.8b between
the µ axis and the µ-v characteristic.
We can see that this energy depends directly on the velocity acquired by the inertor, but not on its
force. Hence the inertor stores energy by virtue of its velocity. The inertor energy is always positive and
does not depend on the sign of either its velocity or acceleration.
Figure 2.9: (a) Symbol of pure source of velocity. (b) Actuator of prescribed relative velocity and (c) its
multipole model.
Fig. 2.9a gives our graphical symbol for the pure source of velocity. The two pins sticking out of
the circle denote the source poles representing the energy interaction of the source with the rest of the
system. Besides the prescribed velocity v(t), the source is associated with its force F (t). This force
counterbalances the forces imposed on the source by the system in which the source operates. The
source polarity denotes the + sign by one of its poles. The assumed positive orientation of v(t) and
F (t) in relation to the symbol polarity is indicated in Fig. 2.9a by the empty-head and full-head arrow,
respectively.
Fig. 2.9b shows a real rectilinear actuator of negligible mass acting between the interaction sites A
and B so that ẋBA (t) = va (t) where va (t) is the actuator velocity. Let us assume that va (t) is sufficiently
independent from the external forces Fext and −Fext exerted on the actuator by the system. Then the
actuator can be modelled by a pure source of velocity as shown in Fig. 2.9c. The rectilinear motion of
the interaction sites A and B in the x-direction is represented in the model by nodes Ax and Bx .
Fa and −Fa are the internal actuator forces counterbalancing the external forces. Let us assume that
the source symbol modeling the actuator is placed in the diagram in such a way that the orientation of
its velocity v is in agreement with the velocity va of the actuator. Then the source force F represents the
internal actuator force Fa acting at the site of interaction represented by the node to which the + pole
of the source is attached.
2.5. PRESCRIBED RECTILINEAR INTERACTIONS 2-9
Example:
Figure 2.10: (a) Car towing, (b) its model. (c) Towing velocity profile, (d) corresponding force exerted
by towing car, (e) energy accumulated in towed car, (f ) position of both cars.
The most trivial dynamic model of the two-car system shown in Fig. 2.10a is given in Fig. 2.10b. As
it is assumed that all the system bodies are rigid and the towing rod is massless a single point – the point
A – is sufficient for representing the system rectilinear motion along x. Only inertia is considered in the
towed car dynamics and the towing car is modelled by a source of velocity.
Let us consider the case when the velocity profile of the towing car velocity vc2 is triangular (Fig. 2.10c).
The source force Fc2 response (Fig. 2.10d) is then negative constant when the velocity increases, and
positive constant when the velocity decreases. Therefore, the towing car supplies energy into the system
in the former case whereas it consumes energy in the latter case. In both cases, the energy accumulated
in the towed car is positive (Fig. 2.10e).
The forces Fc1 and Fc2 are in balance, i.e.
Fc1 (t) + Fc2 (t) = 0 (2.23)
In conformity with our conventions, the arrows indicating orientation of the forces Fc1 and Fc2 oppose the
x-direction in the system geometric representation, while they are directed away from the non-reference
node Ax in the model of the system.
The initial condition of the integration xA (t0 ) corresponds to the initial position of point A at t = t0 .
The car position is for the triangular velocity profile shown in Fig. 2.10f .
In a similar way, a differentiating block can be used to indicate the car acceleration. This illustrates
Fig. 2.11b in which the differentiator is attached to the node Ax . The indicated acceleration
d
ẍA = ẋA
dt
is denoted at the differentiating block output as dAx.
Note that attaching a block input to a node does not disturb the force balance at the node as no force
enters any block.
F = f (t) (2.24)
Figure 2.12: (a) Symbol of pure source of force. (b) Actuator of prescribed force and (c) its multipole
model.
Fig. 2.12b shows a real rectilinear actuator of a prescribed force Fa acting along the x-axis between
the interaction sites A and B of a system. The actuator force Fa is counterbalancing the external force
Fext to the actuator. The effect of the real actuator is modelled in Fig. 2.12c using the pure source of
force. The motion of the interaction sites A and B in the x-direction is represented in the model by the
nodes Ax and Bx .
Polarity of the full-head arrow for the force Fa indicates that if Fa > 0 the actuator pulls the sites A
and B closer. If Fa < 0 the actuator pushes the sites A and B apart.
The source symbol should be oriented between the nodes in agreement with the orientation of the
x axis associated with the real actuator. Therefore, the + pin of the symbol should be attached to the
node Bx representing the actuator site of interaction B that is assumed to move ‘faster’ then its site A. If
2.5. PRESCRIBED RECTILINEAR INTERACTIONS 2-11
the source is oriented properly its force F represents the prescribed force Fp acting at the site B related
to the + pole of the source.
If the source symbol modeling the actuator is placed in the diagram in such a way that the orientation
of its velocity v is in agreement with the velocity va of the actuator, the source force F represents the
actuator force Fa at the actuator end which is related to the + pole of the source.
If the source velocity v and force F are either both positive or both negative the modeled actuator
consumes energy supplied by the remaining system. In the former case, the system forces the actuator to
extend its length whereas in the latter case the actuator length is shortened by the system. If, however,
the signs of the source velocity v and force F are different the modeled actuator delivers energy into the
system. If both signs are positive the actuator pushes the system sites apart, if both signs are negative
the sites pulled closer to each other.
Examples:
The free vertical fall of a ball shown in Fig. 2.14a is modeled in Fig. 2.14b using a source of force Fg
to model the gravitational force Fg = 9.81m acting on the body. The ball air resistance is neglected in
the model.
Figure 2.15: (a) Symbol of ideal indicator of velocity. (b) Indication of a source-of-force velocity.
Pure sources of zero velocity can be utilized as physical elements modeling ideal measuring instruments
or indicators of force. Our graphical symbol for such an indicator is given in Fig. 2.16a. Direction
of the arrow showing the assumed positive polarity of the indicated force is fixed to the polarity of the
2-12 MODULE 2. RECTILINEAR MECHANICAL SYSTEMS
symbol. Fig. 2.16b shows this symbol utilized to indicate the force of a source of velocity representing
the force exerted on a velocity actuator by the system in which it operates.
Note the difference between measuring velocity and force. The direct measurement of forces requires
disconnecting the system and connecting it back by a force indicator. Velocities (as well as positions
or accelerations) are measured by indicators applied to systems without any need to disconnect them.
This is the reason why velocities represent rectilinear across variables while forces are rectilinear
through variables.
Figure 2.16: (a) Symbol of ideal indicator of force. (b) Indication of a source-of-velocity force.
Re-solvable examples
Body driven by given force
Body driven by given velocity
Vertical throw of a ball
Body driven by a frequency-variable force
Figure 2.17: (a) Symbol of pure damper, (b) F − v characteristic of the ideal damper.
Fig. 2.17a gives our graphical symbol for pure rectilinear dampers. A pure rectilinear damper
exerts a resistive force F that is a function of the relative velocity v of its motion, but exhibits no mass
or spring effects. Thus,
F = f (v) (2.25)
where f (·) is a function such that F = 0 when v = 0. In the case of an ideal rectilinear damper this
constitutive relation is linear, i.e.,
F (t) = b · ẋ(t) (2.26)
where b [N.s/m] is the damping constant of the ideal damper. The ideal damper characteristic is shown
in Fig. 2.17b.
Characteristics close to those of an ideal damper demonstrate, for example, the sliding surfaces shown
in Fig. 2.18a if they are separated by a thin layer of a fluid lubricant, or the dashpot in Fig. 2.18c if it
does not move too fast. In such cases the friction forces result from the viscous drag of a fluid under
laminar flow conditions.
2.6. RECTILINEAR DAMPING 2-13
Fig. 2.18a shows sliding surfaces in contact pushed towards each other by some forces Fn normal to
the motion. The frictional forces Fr resisting a relative motion of the surfaces must be compensated for
by external forces Fext .
Figure 2.18: (a) Sliding surfaces in contact, (b) corresponding model. (c) Oil-filled shock absorber.
In Fig. 2.18b, the friction effect between sliding surfaces is modelled by an ideal damper. Nodes Ax
and Bx in the model represent x-motions of points A and B of the real surfaces in the x-direction. If the
surfaces are well lubricated by a fluid, the damping constant
Aν
b= (2.27)
d
where A is the common area and d is the thickness of the lubrication film between the surfaces, ν is the
viscosity coefficient of the lubricant.
Note the way in which the polarities of the damper relative velocity v and resistive force F are
associated with the damper symbol shown in Fig. 2.17a. As the symbol is asymmetric, no + sign is
necessary to denote the symbol polarity (as it was the case of pure sources). If the damper symbol is
placed in a multipole diagram in such a way that its velocity v represents polarity of the relative velocity
ẋBA , then the damper force F represents the resistive force Fr at the ‘faster’ site of interaction. The
damper force F thus corresponds in Fig. 2.18b to the resistive force counterbalancing the net external
force Fext acting on point B in the real bodies.
Since no mass is associated with the surfaces, the external forces Fext applied to the interaction sites
A and B are balanced, and the same must be true about the reaction forces Fr at these sites. Thus an
external force applied to point B is assumed to flow through the damper to point A without any change.
The shock absorber shown in Fig. 2.18c consists of a piston inside an oil-filled cylinder in which fluid
is forced through a narrow orifice in the piston by relative motion, thus producing resisting force. In this
case, the damping constant can be approximated as
" 2 #2
D
b = 8πνℓ −1 (2.28)
d
where D is the cylinder diameter, d is the diameter of the orifice, and ℓ is the length of the orifice.
The mechanical power dissipated in the pure damper as heat is
P = ẋ · F
Note that it is always positive regardless of the direction of the damper relative motion.
Re-solvable examples
Viscous friction experiment
Two-mass model of car
2-14 MODULE 2. RECTILINEAR MECHANICAL SYSTEMS
Figure 2.19: (a) Symbol of pure spring. (b) Coil spring stretched and compressed. (c) Ideal-spring
deformation characteristic.
The spring deformations due to compliance are often approximated to be proportional to the ap-
plied force. A linear pure spring, which is called an ideal spring, has a proportional relation between
deformation and force,
∆ℓ = L · F (2.29)
where ∆ℓ [m] is the spring deformation, F [N] is the spring restoring force, and the constant parameter L
[m.N−1 ] is the compliance factor of the spring. The reciprocal value of the compliance factor, K = 1/L
[N.m−1 ], is the spring stiffness termed also the spring constant.
For example, the compliance factor of a coil spring can be derived using the methods of applied
mechanics to be approximately
8D3 n
L=
Gd4
where D [m] is the spring pitch diameter (mean coil diameter), d [m] is the diameter of the wire, n [–] is
the number of coil turns, and G [N.m−2 ] is the shear modulus of the coil material.
The compliance factor of a cylindrical bar made of homogenous ideally elastic material is according
to Hook’s law
ℓ0
L= (2.30)
E ·A
where ℓ0 [m] is the force-free length of the rod, A [m2 ] is the rod cross-sectional area, and E [N.m−2 ] is
the Young’s modulus of elasticity of the rod material.
Fig. 2.19b shows a stretched, force-free and compressed spring. The spring deformation ∆ℓ = ℓ − ℓ0
where ℓ is the spring actual length ℓ = xB − xA and ℓ0 is the free length of the spring. If ∆ℓ > 0 the
spring is stretched, if ∆ℓ < 0 the spring is compressed. Fig. 2.19c shows the ℓ − F characteristic of the
ideal spring.
By differentiating (2.29), we can arrive at the constitutive relation of an ideal rectilinear springs
d
v =L· F (2.31)
dt
where v = dℓ/dt is the spring relative velocity. Alternatively, the relation (2.31) can be expressed in the
integral form
Z t
F =K v dτ + F (t0 ) (2.32)
t0
Fig. 2.20a shows a coil spring placed between the interaction sites A and B in the x-motion. If the
spring mass is negligible in comparison with masses of the interacting components it can assumed that
2.7. RECTILINEAR COMPLIANCE 2-15
Figure 2.20: (a) Coil spring, (b) force-controlled and (c) elongation-controlled model. (d) Model respecting
spring mass and damping. (e) Distributed-parameter model.
the external forces Fext applied to the spring end-points are balanced. This implies that also the restoring
forces Fr , opposing the external forces, are balanced at the spring end-points.
Note that one of the pins of the pure-spring symbol given in Fig. 2.19b is denoted by the + sign.
This allows for an unambiguous association of the polarities of the spring velocity v and the spring force
F with the polarity of the symbol. In Fig. 2.20b, the pure-spring symbol is oriented in conformity with
the chosen orientation of the modelled real spring shown in Fig. 2.20a. Then the pure-spring force F
represents the real-spring restoring force Fr at the assumed ‘faster’ end of the real spring.
To express the total length of the modelled spring, the pure spring is combined with an integrator in
Fig. 2.20b. For a chosen pre-loading force F (t0 ) of the spring, the initial relative distance between the
spring end-points at t = t0 should be specified as
The source is controlled by the relative distance of the spring end points xBA . The initial value of this
distance for a chosen initial length of the spring ℓ(t0 ) is
Figure 2.21: (a) Suspended ball, (b) its dynamic model, (c) ball position.
Examples:
Fig. 2.21b shows a dynamic model of the ball suspended on a spring given in Fig. 2.21a. Node Ay
represents the y-motion of point A of the ball. The spring symbol is placed between the node Ay and
the reference node in such a way that the polarity of its velocity v is in agreement with the polarity of
the relative velocity of the real spring along the y axis. As a result, the pure spring force F represents,
2-16 MODULE 2. RECTILINEAR MECHANICAL SYSTEMS
both by its magnitude and polarity, the actual restoring force counterbalancing the gravitational force Fg
acting on point A in the real bodies. The effect of the gravitational force Fg = 9.81m is modelled using
a source of force.
Figure 2.22: (a) Car following bumpy road, (b) car-wheel suspension, (c) quarter-car model.
Fig. 2.22a shows a car following a bumpy road while Fig. 2.22b shows just a detail of the car, the
suspension of one of its wheels. In Fig. 2.22c point Ry represents the y-motion of the contact point
between the car tyre and the road surface R when the car moves in the x-direction. Points Wy and By
represent the y-motion of the wheel axis W and the body-suspension interaction point B, respectively.
Lt and bt is the compliance and damping of the wheel tyre, respectively. Similarly, Ls and bs is the
compliance and damping of the wheel tyre, mw is the mass of the complete wheel and mb is the quarter
of the mass of the car body. The model is excited by the source of the velocity ẏR resulting from the tyre
copying the road bumps.
2.8 Summary
To set up a dynamic model of a real rectilinear system and to simulate its behavior using DYNAST, you
should proceed through the following steps.
2.8. SUMMARY 2-17
Fig. 3.1 shows a subsystem with two concentric shafts rotating about the z-axis fixed to the reference
frame. The subsystem might be a torsional spring, a rotational damper, or a gearbox, for example. The
ends of the shafts A and B are in contact with shafts of some other subsystems. The power intake of the
subsystem due to the rotational motion of its energy entries A and B about the z axis is
where ϕ̇A (t) and ϕ̇B (t) are absolute angular velocities of the entries A and B about the z axis. τA (t) and
τB (t) internal torques counterbalancing external torques τA ext(t) and τB ext(t) applied to the entries.
Absolute angular motion about the axis z is described in Fig. 3.1 by the angular displacement
ϕ, i.e., by the angle subtended between a line drawn on the rotating entry which lies in the x-y-plane
perpendicular to the axis of rotation z and the coordinate axis x fixed to the reference frame. In this
text, the counterclockwise direction will be considered as the positive orientation of rotation in a plane
perpendicular to the axis of rotation.
Absolute angular velocity ϕ̇(t) [rad/s] and absolute angular acceleration ϕ̈(t) [rad/s2] of an
entry is by definition
dϕ dϕ̇ d2ϕ
ϕ̇(t) = and ϕ̈(t) = =
dt dt dt2
where ϕ(t) is the angular displacement of the entry. The relative angular displacement, relative
angular velocity and relative angular acceleration of the entry B with respect to the entry A is
3-1
3-2 MODULE 3. ROTATIONAL MECHANICAL SYSTEMS
3.1.2 Torque
A moment of force or torque τ [N.m] can be viewed as the time rate of the angular momentum
ρ(t) acting in the same direction, i.e.
dρ
τ=
dt
The complete specification of a moment of force τ (t) must include its magnitude and the axis of rotation.
To have the energy flow through an entry into a subsystem positive, the torques acting on the entry must
be oriented in the opposite way with respect to the direction of the entry angular velocity.
Figure 3.2: (a) Symbol of rotational inertor. (b) Flywheel, and (c) its model.
Rotational inertors model inertial effect in rotational motion of bodies exclusively, they ignore any
damping or spring effects as shown in Fig. 3.2. The angular velocity ϕ̇(t) of a rotational inertor represents
the absolute angular velocity of a body measured in relation to a nonaccelerating reference frame. Thus
the −pole of rotational inertors must be always coalesced with a node corresponding to the reference.
This is the reason why only the +pole is shown in the element symbol.
If the moment of inertia JT of a body about a fixed axis passing through the mass center T is known,
the moment of inertia J of the body about another fixed axis parallel to the former one can be determined
as
J = JT + md2 (3.3)
where m is the mass of the body, and d is the separation of the axes.This formula, known as the parallel-
axis or Steiner’s theorem, is illustrated by Fig. 3.3. The diameter of the ball of the mathematical
pendulum of Fig. 3.3a is so small that its moment of inertia is negligible. If also the mass of the pendulum
string is neglected, the pendulum moment of inertia with respect to the axis of rotation A is J = mr2 .
Fig. 3.3b shows a compound pendulum. If only the moment of inertia JT with respect to mass center
of the compound pendulum body is known, the pendulum moment of inertia with respect to the axis of
rotation A must be determined using (3.3). The mass center of each of the pendulums is acted on by the
moment of the gravitational force Fg = mg, the moment arm of length r sin ϕ varies with the pendulum
deflection ϕ. In the dynamic model for both pendulums shown in Fig. 3.3c, the ideal rotational damper
of factor B respects damping of the pendulums.
Figure 3.3: (a) Mathematical and (b) compound pendulum, (c) pendulum dynamic model.
where τ (t) is the moment of force, ϕ̇(t) is the relative angular velocity, and br [N.m.s/rad] is the rotational
damping constant of the damper.
Figure 3.4: (a) Symbol of pure rotational damper. (b) Oil-filled rotational damper, and (c) its model.
An ideal rotational damper is a twopole model that ignores any mass or spring effects and assumes
that a moment of force is proportional to the relative angular velocity of its poles. An example of a real
device with the τ -ϕ̇ characteristic close to that of the ideal damper, is the viscous rotational damper
shown in Fig. 3.4b. It is a propeller rotating in oil inside a cylinder and producing the resisting moment
of force.
The poles represent ϕ-motions of the energy entries B and A of a real damping subsystem. τ represents
the internal moment of force at the entry associated with the +pole, i.e. at the entry B, counterbalancing
an external moment of force applied to this entry. The positive direction of τ is opposite to that for the
damper angular velocity ϕ̇ corresponding to the relative velocity of the entries ϕ̇BA = ϕ̇B − ϕ̇A .
Since an ideal rotational damper has no moment of inertia, the moments of force acting on it must be
balanced. Thus, an external moment of force applied to Bϕ in the ϕ-direction flows through the damper
and out through Aϕ without a change and, at the same time, the internal moment of force at the damper
−pole denoted by Aϕ must be −τ . Note that the power flow into the ideal damper P = ϕ̇τ is always
positive regardless of the direction of its relative motion.
Fig. 3.5a shows the symbol of the ideal rotational damper used to model a shaft passing through a
well lubricated journal bearing fixed to the reference frame. In Fig. 3.5b, the same model is utilized for
an oil-polluted clutch the discs of which are slipping with respect to each other so that they rotate with
different angular velocity and there is a viscous friction effect between them.
Figure 3.5: Ideal model of (a) a shaft in journal bearing, and (b) a slipping clutch.
3-4 MODULE 3. ROTATIONAL MECHANICAL SYSTEMS
or
d
ϕ̇(t) = Lr τ (t) (3.6)
dt
where ϕ(t) is the ideal spring angular displacement, ϕ̇(t) is the ideal spring angular velocity, τ (t) is
the ideal spring moment of force, and Lr [rad.N−1 .m−1 ] is the torsional compliance of the spring. The
reciprocal value of Lr is the spring torsional stiffness or torsional spring constant.
Figure 3.6: (a) Symbol of pure torsional spring. (b) Spiral torsional spring, and (c) its model. (d)
Helical-coil spring. (e) Twisted rod.
The pure spring has no inertia by definition and hence transmits torque undiminished. One of the most
common mechanical components which exhibit the characteristics of a rotational spring is the torsional
spiral spring shown in Fig. 3.6b. Fig. 3.6c shows a helical-coil torsional spring, and Fig. 3.6d a twisted
rod. ϕBA corresponds to the relative angular deformation of the real spring, while τ represents the spring
internal moment of the force counterbalancing the external moment of a force applied to the spring.
Figure 3.7: (a), (b) torsional pendulum. (c), (d) double torsional pendulum.
Module 4
Motion of point A with respect to point A along the x axis is represented by poles Ax and A x ,
respectively. Similarly, motion of point B with respect to point B along the ξ axis is represented by
poles Aξ and A ξ . Fx is the force exerted by the transformer to move points A and A apart while
counterbalancing external forces applied to these points. Fξ is the force counterbalancing external forces
applied to points B and B .
The relation between velocities associated with the translational transformer is defined as
ẋ/ξ˙ = n (4.1)
where ẋ is the relative velocity of point A with respect to point A , and ξ̇ is the relative velocity of point
B with respect to point B , n is the transformer ratio. An ideal transformer has this ratio constant.
Forces associated with the translational transformer are related by definition as
Fξ /Fx = −n (4.2)
P = Fx ẋ + Fξ ξ˙ (4.3)
After submitting (4.1) and (4.2), (4.3) becomes P = 0. A pure translational transformer thus transforms
a translational motion into another translational motion and vice versa. It performs this function with
no energy loss or storage, so that the net power flow into a pure transformer is zero.
4-1
4-2 MODULE 4. COUPLED MECHANICAL SYSTEMS
a
b
a
−
b
−1
ra
−
rb
Table ?? shows examples of pure models of some translatory-to-translatory transducers. They include
levers, pulleys and a rigid link of a fixed length the endpoints of which are constrained to a linear or
circular path.
4.1.2 Levers
A translatory-to-translatory coupling is typified by a lever which has no mass, is infinitely stiff, and is
free from frictional effects. Such a lever is shown in Fig. 4.2a.
The relationship between the velocities of endpoints A and B of the lever with respect to the pivot C
is
ẏA /ẏB = −a/b = n (4.4)
where n is the lever ratio. Note that a and b change as ϕ varies, but that a/b remains constant. Often
ϕ will be limited to a small angle, so that a and b are essentially constant.
The full-head arrows denote forces FA , FB and FC exerted by the lever at its end points A and B and
at the pivot C, respectively, to counterbalance external forces applied to these points. Since the massless
lever has no momentum, then by definition the sum of the forces in any direction and the moments of
the forces about the pivot must be zero:
FA + FB + FC = 0 (4.5)
4.2. ROTARY-TO-ROTARY COUPLINGS 4-3
FA a + FB b = 0 (4.6)
Therefore,
FB /FA = a/b = −1/n (4.7)
Note, that the relationships between velocities (4.4) and between forces (4.7) are the same as in the
translational transformer used in Fig. 4.2b to model the lever.
Rotational motion of point A with respect to point A rotating about the same axis with angular
velocity ϕ̇ is represented by poles Aϕ and A ϕ , respectively. Similarly, the rotational motion of point
B with respect to rotating point B with the velocity ψ̇ is represented by poles Aψ and A ψ . τϕ is the
torque exerted by the transformer to move points A and A apart while counterbalancing external torques
applied to these points. τψ is the force counterbalancing external torques applied to points B and B .
Relation between angular velocities associated with the rotational transformer is
ϕ̇/ψ̇ = n (4.8)
where ϕ̇ is the relative angular velocity of point A with respect to point A , and ψ̇ is the relative angular
velocity of point B with respect to point B . n is the transformer ratio. An ideal transformer has this
ratio constant.
Torques associated with the rotational transformer are related as
τψ /τϕ = −n (4.9)
P = τϕ ϕ̇ + τψ ψ̇ (4.10)
provided that the gears are rigid and the gear teeth are geometrically perfect. The radii ra and rb are
called the pitch radii of the gears, and their magnitudes are proportional to the numbers of teeth Na and
Nb . The gear ratio is defined as
ra /rb = Na /Nb
tA tA jA
jA tA tB
tB
tB jA jB
jB
jB
(a) (b) (c)
Figure 4.4: (a) Gear train, (b) its free-body diagram, (c) transformer model.
Since the gears are assumed to roll without slipping, the arc distance ra ϕ̇A ∆t turned off on gear a
must equal minus the arc distance rb ϕ̇B ∆t turned off on gear b. Hence
ϕ̇A rb Nb
=− =− =n (4.11)
ϕ̇B ra Na
where the negative sign indicates that the direction of ϕ̇B is opposite to that of ϕ̇A .
To determine the relationship between the torques τA and τB , let us isolate the two gears from each
other and from the frame, thus forming their free-body diagrams of Fig. 4.4b. Only torques and torque-
producing forces are shown on these free bodies, and friction between the gear-tooth surfaces is assumed
to be zero. If we further assume that the gears have zero inertia or constant speed, the net torque on
each gear is zero. If we also assume the bearing torques to be negligible,
τA = Ft ra and τB = Ft rb
Employing now the definition of n in (4.11), we obtain
τA ra Na
= = = −1/n (4.12)
τB rb Nb
Thus the speed is changed as n while the torque is changed as 1/n. The net power into the gear train
for ϕ̇A = 0 is therefore zero:
P = τA ϕ̇A + τB ϕ̇B = τA ϕ̇A + (τA /n)(−τA n) = 0
Since the gear system is assumed to have constant angular momentum, the torque τC shown acting
on the container Fig. 4.4a must be equal to minus the sum of τA and τB , or
τA + τB + τC = 0 (4.13)
Since ϕ̇C = 0 in the preceding discussion, τC does no work on the system.
Table 4.2 shows various setups of gear trains that can be modeled by a single pure rotational trans-
former. The table gives the corresponding gear ratios.
External rb
spur gears −
ra
Internal rb
spur gears
ra
Bevel rb
gear pair
ra
Planet rb
gear
ra
Skew rb
gear pair
ra
4-6 MODULE 4. COUPLED MECHANICAL SYSTEMS
tB tA tB
jB
tA jA
jAB jAB jA t(jAB) j
B
Figure 4.5: (a) Gear train with backslash, (b) backslash characteristic, (c) multipole model.
tB tB
jA
tA
jA
jB jB
tA
(a) (b)
Fig. 4.6 shows pure belt-and-pulley and chain-and-sprocket systems that perform essentially the same
function as gear trains. As opposed to a belt and pulley, the latter system does not allow slippage between
the belt and sprocket due to the sprocket teeth. Note that the sign of n is positive in the case of the
system in Fig. 4.6a as the pulleys rotate in the same direction, whereas in the case of the system in
Fig. 4.6b the sign is negative.
The Eytelwein’s formula gives the relation between the forces FA and FB acting at the belt-segment
endpoints A and B. For the case when the belt is just about to slip if pulled to the right, i.e. if FA > FB > 0,
4.2. ROTARY-TO-ROTARY COUPLINGS 4-7
The function sgn(Ft ) was included into the formula to make it valid also for the case when the belt
is pulled to the left, i.e. for FB > FA > 0. (You are encouraged to check its correctness.)
Equation (4.16) can be used both for static friction of impending slipping as well as for the slipping
kinetic friction after substituting there either the coefficient of static friction µs or the coefficient of kinetic
friction µk for µ. Fig. 4.8 shows an example of the Ff -ẋ characteristic of combined static and kinetic
friction governed by (4.16) for the tension Ft = ±1. Note that the characteristic is asymmetric with
respect to the origin.
As shown in Fig. 4.7d, the belt friction model from Fig. 4.7c can be completed to respect the belt
segment flexibility kb , and mass mb . The belt friction model we have derived for moving belt and fixed
drum applies equally well to problems involving fixed belt and rotating drum, like band brakes, and to
problems involving both belt and drum rotating, like belt drives. It can be also used to systems with the
belt replaced by a rope or a string.
In case of V-shaped belts (4.16) should be replaced by
Ff = Ft 1 + eµγ·sgn(Ft )/ sin(δ/2)
(4.17)
where δ is the angle of the ‘V’ of the belt cross section.
Fig. 4.9a shows an example of a hoisting machine lifting a load by means of a cable passing over an
undriven sheave and wrapped around a driven drum. The rotary motion of the drum drive is transformed
into the translational motion of the cable by means of an ideal transformer with the ratio 1 : r1 , where r1
is the drum radius. The c1 and c2 parts of the cable are modelled by elements characterizing the cable
flexibility and damping. The element parameters of the cable part c2 depend on the cable length ℓ of
c2 . The friction between the cable and the sheave is represented by the belt-friction model given above.
The sheave dynamics, converted from rotary to translational motion along the cable, is represented by
the elements of reduced parameters
mr = Js /r2 , br = bs /r22 (4.18)
where Js is the moment of inertia of the sheave, r2 is its radius and bs is the damping of the sheave
bearing.
4-8 MODULE 4. COUPLED MECHANICAL SYSTEMS
B y
c1
g
C A 0
w
G
F c2
E -l (a)
b1 b2(l) l0
l
F E
DRIVE G k1 B BELT A k (l)
w 1:r1 FRICTION
2 mg
& DRUM
m
C
br
mr (b)
Rotational motion of point A with respect to point A along the x axis is represented by poles Ax
and A x , respectively. Rotational motion of point B with respect to point B rotating about the same
axis with the velocity ϕ̇ is represented by poles Aϕ and A ϕ . F is the force exerted by the transformer
to move points A and A apart while counterbalancing external forces applied to these points. τ is the
torque counterbalancing external torques applied to points B and B .
Relation between angular velocities associated with the rotational transformer is
ẋ/ϕ̇ = n (4.19)
where ẋ is the relative velocity of point A with respect to point A , and ϕ̇ is the relative angular velocity
of point B with respect to point B . n is the transformer ratio. An ideal transformer has this ratio
constant.
4.3. TRANSLATORY-TO-ROTARY COUPLINGS 4-9
The force and torque associated with the translatory-rotary transformer are related as
F/τ = −n (4.20)
P = F ẋ + τ ϕ̇ (4.21)
After submitting (4.19) and (4.20), (4.21) becomes P = 0. A pure translatory-rotary transformer
is thus an idealized mechanical device which transforms a translational motion into a rotational motion
and vice versa. A pure transformer performs this function with no energy loss or storage, so that the net
power flow into a pure transformer is zero.
4.3.2 Rack-and-pinion
In motion control problems it is often necessary to convert rotational motion into a translational one and
vice versa. For instance, a load may be controlled to move along a straight line through a rotary motor
and rack-and-pinion assembly, such as that shown in Fig. 4.11a. The pinion is a small gear, and the rack
is a linear member with the gear on one side. The describing relation of this system is defined as
x = rϕ (4.22)
tA tA
jA jA tA FB
jA xB
FB FB
Figure 4.11: (a) Rack-and-pinion gear train, (b) equivalent system, (c) pure model.
tA jA tA FA
FA
jA FB xA
FB
xB
xA xB
(a) (b)
To apply this result to the rack-and-pinion assembly movable along the x-axis as that shown in
Fig. 4.12a, the variable x should be replaced by xAB .
4-10 MODULE 4. COUPLED MECHANICAL SYSTEMS
FB
FB tA FB
tA jA
tA
jA jA xB
Figure 4.13: (a) Pulley or sprocket. (b) Belt and pulley rotary-translatory assembly.
ϕ = ±P x
n = ±P
The sign is related to the mutual orientation of the rotational and translational motions.
tA FB
jA
jA xB
tA
(a) (b)
tA FB
tA
jA xB
FB
(a) (b)
As the crank rotates by 360◦, the linear stage moves in both directions along x. There are two
distinct angular positions of the crank that correspond to full extension and retraction of the stage called
dead-center positions. If this module is used to convert linear to rotary motion the direction of rotation
may go in either way in these positions (this problem is usually overcome by combining the crank with
a flywheel).
C
E E
F F
D (a ) (b )
Figure 4.16: (a) Cascade interconnection of energy-transfer transformers, (b) one-transformer equivalent.
Fig. 4.16a shows a ‘cascade’ interconnection of energy-transfer transformers of coupling ratios n1 and
n2 . A simple analysis of the total transformation of across and through variables in the interconnection
indicates that the cascade can be replaced by only one transformer with the coupling ratio n1 n2 shown
in Fig. 4.16b.
Fig. 4.17 indicates the way in which series or parallel combinations of physical elements connected
to a pole-pair of an energy-transfer transformer are ‘seen through’ the transformer, or referred to at the
other transformer pole-pair. Z resp. Y is an impedance resp. admittance of the generic element, E resp.
J is an independent source of an across resp. through variable. This is the principle behind utilizing
various transducers and couplings to convert the amount of an actuator load at the price of changed
related across and through variables.
4-12 MODULE 4. COUPLED MECHANICAL SYSTEMS
Module 5
Figure 5.1: (a) Particle position, (b) velocity, and (c) acceleration vectors.
The particle position can be completely specified by the position vector r measured from a fixed origin
O. At time t + ∆t, the particle is at point A’, the position of which is specified by the vector r + ∆r.
The particle displacement during time ∆t is indicated by the vector ∆r representing the vector change
of position. If an origin O’ is chosen at some location different from O, the position vector r changes, but
∆r remains unchanged. The distance traveled by the particle as it moves along the path from A to A’ is
the scalar length ∆s measured along the curved path. Thus, the vector displacement ∆r is distinguished
from the scalar distance ∆s.
The instantaneous velocity v of the particle is defined as the limiting value of the particle average
velocity between the points A and A’ as the points become closer together and the time interval ∆t
approaches zero. Thus,
∆r
v = lim
∆t→0 ∆t
The direction of ∆r approaches that of the tangent to the path as ∆t approaches zero and, hence, the
velocity v is always a vector tangent to the path as shown in Fig. 5.1b.
The definition of the derivative of a scalar quantity can be extended to include a vector quantity so
that
dr
v= = ṙ
dt
Thus the derivative of a vector is itself a vector having both a magnitude and a direction. The magnitude
of v is called the speed and is the scalar
ds
v = |ṙ| = = ṡ
dt
5-1
5-2 MODULE 5. PLANAR MECHANICAL SYSTEMS
In Fig. 5.1b, the velocity of the particle at A is denoted by the tangent vector v and the velocity at A’
by the tangent vector v′ . There is thus a vector change in the velocity during the time ∆t. The velocity
v at A plus the change ∆v must equal the velocity at A’. So we may write v′ − v = ∆v. Inspection of
the vector diagram in Fig. 5.1c shows that ∆v depends both on the change in magnitude of v and on the
change in direction of v.
The instantaneous acceleration a of the particle is defined as the limiting value of the average accel-
eration between A and A’ as ∆t approaches zero, or
∆v
a = lim
∆t→0 ∆t
By definition of the derivative we write
dv
a= = v̇ = r̈
dt
As the interval ∆t approaches zero, the direction of the change ∆v approaches that of the differential
change dv and, hence, a. It is apparent from Fig. 5.1c that the acceleration a reflects change in magnitude
of v as well as the change of direction of v. In general, the direction of the acceleration of a particle in
curvilinear motion is neither tangent nor normal to the particle path.
Figure 5.2: Rectangular components of (a) of position, velocity, and (b) acceleration.
The position vector rA , the velocity vA , and the acceleration aA of the particle are represented in
Fig. 5.2 together with their x- and y-components. Using the unit vectors i and j, we may write the vectors
rA , vA , and aA as
rA = xA i + yA j
vA = ṙA = vAx i + vAy j (5.1)
aA = v̇A = r̈A = aAx i + aAy j
The unit vectors have no time derivatives since their magnitudes and directions remain constant. Thus
the scalar values of the components of vA and aA are just vAx = ẋA , vAy = ẏA and aAx = v̇Ax = ẍA ,
aAy = v̇Ay = ÿA .
It is obvious from Fig. 5.2 that the absolute values vA = |vA | and a = |aA | are related to the x- and
y-components as q q
vA = 2 + v2
vAx aA = a2Ax + a2Ay
Ay
Relations (5.1) prove that the planar motion of a particle can be decomposed into motions in two
different directions (mutually perpendicular in this case). Practical experience indicates that the opposite
procedure – vector composition of planar motions – is also possible. Such principle of motion composition
is considered as an axiom, it cannot be proven.
Thus, if the coordinates x and y of a particle are known as functions of time, x = fx (t) and y = fy (t),
then for any value of the time we may combine them to obtain r. Similarly, we combine their first
5.1. PLANAR MOTION OF PARTICLES 5-3
derivatives ẋ and ẏ to obtain v and their second derivatives ẍ and ÿ to obtain a. Or, if the acceleration
components ax and ay are given as functions of the time, we may integrate each one separately with
respect to time, once to obtain vx vy and again to obtain x = fx (t) and y = fy (t). Elimination of the
time t between these last two parametric equations gives the equation of the curved path y = f (x).
Figure 5.3: Relative motion (a) of A with respect to B, and (b) of B with respect to A.
Using the unit vectors i and j along the x- and y-axes we can write
where xAB = xA − xB and yAB = yA − yB are the x- and y-components of the relative position of A with
respect to B.
If we now differentiate (5.2) and (5.3) with respect to time we obtain the relative velocity of A as
observed from B
vAB = ṙAB = ṙA − ṙB
where vABx = vAx − vBx and vABy = vAy − vBy are the x- and y-components of the relative velocity of A
with respect to that of B. These can be expressed in terms of the rectangular components of the absolute
velocities vAx , vBx , vAx , vBy of points A and B.
If we differentiate (5.2) and (5.3) with respect to time twice we will get the relative acceleration of A
with respect to B as
aAB = ȧAB = ṙA − ȧB = aABx i + aABy j
where the x- and y-components aABx = aAx − aBx and aABy = aAy − aBy .
Let us now exchange the role of the two points to observe the motion of B as seen from A. Fig. 5.3b
indicates that the corresponding expressions for the relative position, velocity, and acceleration become
rBA = rB − rA = −rAB
vBA = vB − vA = −vAB
aBA = aB − aA = −aAB
or
(xAB + xBC + xCA )i + (yAB + yBC + yCA )j = 0
when the vectors are decomposed into their x- and y-components. Thus postulate of motion compatibility
applies independently also to the x- and y-components
The postulate of motion compatibility holds for velocities and accelerations of the three points:
In general, the postulate of motion compatibility holds for any motions along a closed loop passing
through the sequence of an arbitrary number of points. This trivial-looking statement plays an important
role in the formulation of equations describing the behavior of mechanical systems.
In addition to the need for combining forces to obtain their resultant, we often have occasion to
replace a force by its vector components which act in specified directions. By definition, the two or more
vector components of a given vector must vectorially add to yield the given vector. Thus, the force R
in Fig. 5.5 may be replaced by or resolved into two vector components F1 and F2 with the specified
directions merely by completing the parallelogram as shown to obtain the magnitudes of F1 and F2 .
Let us now consider a particle subjected to the action of concurrent forces F1 , F2 , F3 , ... assuming
that the lines of action of all the forces are situated in the same plane as the trajectory of the particle.
The vector sum of the applied forces becomes
X
Fa = Fi (5.4)
i
The sum must include all the external forces applied to the particle in question. The reliable way to
account accurately and consistently for every force is to form the free-body diagram of the particle under
5.1. PLANAR MOTION OF PARTICLES 5-5
consideration isolating it from all contacting and influencing bodies and replacing the bodies removed by
the forces they exert on the particle isolated.
Fig. 5.6a gives an example of three forces applied to a particle represented by point A while Fig. 5.6b
shows the resultant Fa of the applied forces.
Figure 5.6: (a) Forces applied to a particle, (b) vector and (c) orthogonal components of the resultant
external force.
The vector of the total applied force Fa is shown again in Fig. 5.6c where it is decomposed along the
x and y coordinate axes. With the aid of the unit vectors i and j, we may rewrite (5.4) in terms of the
x- and y-components of the individual applied forces as
Fa = Fax i + Fay j
where X X
Fax = Fxi and Fay = Fyi
i i
are the x- and y-components of the resultant force applied to the particle.
The absolute value Fa = |Fa | is
sX X
Fa = ( Fxi )2 + ( Fyi )2
i i
Fa = ma (5.5)
where the acceleration a of the particle is always in the direction of the applied force Fa . This is shown
in Fig. 5.7a where the particle is represented by point A. This effect is the inertia of the particle mass
which is its resistance to rate of change of velocity. Equation (5.5) is usually referred to as the equation
of motion in the vector form.
Figure 5.7: (a) Particle acceleration due to applied force, (b) force dynamic equilibrium.
Fig. 5.7b suggests that the inertial effect of the particle mass can be replaced by the hypothetical
inertial force Fm = −ma acting on the massless point representing the particle. Then (5.5) can be
written in the alternative form
Fa + Fm = 0 (5.6)
interpreted as the dynamic equilibrium of forces acting at on the particle.
5-6 MODULE 5. PLANAR MECHANICAL SYSTEMS
Relation (5.6), known as d’Alembert’s principle, has far-reaching implications in dynamic analysis. It
allows us to treat the particle inertial force Fm in the same way as the external forces applied to the
particle.
If a particle is subjected to the action of a sum of applied forces Fa and its velocity with respect to
the inertial system is zero or constant
Fa = 0
This expresses the static equilibrium of forces acting at on the particle.
Figure 5.8: (a) Submodel for particle motion in the vertical plane, (b) submodel symbol.
Fig. 5.8b gives the three-pole dynamic model of a particle pursuing a curvilinear motion in the vertical
x-y plane which is shown in Fig. 5.8a. The particle position in the x-y plane denotes point Ax. Poles
Ax and Ay represent point velocities in the horizontal and vertical directions. The inertia effects of
the particle mass m in the x- and y-directions are modelled by inertors. A source of force represents
the gravitational force Fg = g m acting on the particle in the vertical direction, g is the gravitational
acceleration.
The third pole corresponds to the absolute frame representing the velocity reference. Fig. 5.8c shows
the symbol of the submodel with pins corresponding to the poles Ax and Ay. The third pole associated
with the velocity reference and respected in the submodel does not need to be shown explicitly in the
symbol.
Obviously, also a multipole model of a particle in a curvilinear motion can be decomposed into two
rectilinear models, one for dynamic interactions in the x direction, and the other one in the y direction.
In most practical cases these two models will be coupled, however.
Figure 5.9: (a) Trajectory of a cannon ball, (b) ball dynamic model.
To model the ball dynamics, we apply the submodel shown in Fig. 5.8 augmented by blocks for
computing the ball xB and xB position by integrating the ball velocities
√ Bx and By . The√ x- and y-
components of the ball initial speed v(0) are obviously vx (0) = v(0)/ 2 and vy (0) = v(0)/ 2.
5.2. PLANAR SYSTEMS WITH MASSLESS RODS 5-7
We can see that the ball x- and y-motions are mutually independent under the simplifying consider-
ation. If we had introduced, for example, aerodynamic drag force exerted on the ball which depends on
the speed squared, then the x- and y-parts of the ball model would be coupled. As a result, the ball x-
and y-motions would be interdependent, and the trajectory would not be parabolic any more.
x2AB + yAB
2
= ℓ2 (5.7)
where ℓ is the rod length whereas xAB = xA − xB and yAB = yA − yB are projections of the rod onto the
x and y axes. After differentiating (5.7) with respect to time we get the geometric constraint imposed by
the rod on the particle motion
xAB ẋAB + yAB ẏAB = 0 (5.8)
Figure 5.10: (a) Particles interconnected by massless rod, (b) transformer model.
The rod must be in equilibrium, i.e. all forces and moments applied to it must be in balance. The
free-body diagram of the rod in Fig. 5.10a indicates that the balance of forces can be written in the form
After substituting for FAx and FAy from (5.9), the previous relation becomes
where ℓ is the rod length. Differentiation with respect to time gives the velocity constrains
B n1
C xB
x m1 By
0 xB xC n2 yB
y C2
m1 m 1g
Cx xC
Cy
yC
m2
m2 m 2g
(a) (b)
Figure 5.13: (a) Mathematical double pendulum, (b) its dynamic model.
Figure 5.14: (a) Carriage with pendulum, (b) its dynamic model.
where
nx = xAB /ρAB ny = yAB /ρAB
Obviously, the resulting model can be composed from two pure transformers as shown in Fig. 5.19b.
where ρ is the distance of A from B. Differentiation with respect to time gives the velocity constrains
Figure 5.17: Massless rod in polar coordinates: (a) graphical symbol, (b) model diagram.
and
FAx cos ϕ + FAy sin ϕ = 0 (5.27)
Balance of moments is
τ − FAx yAB + FAy xAB = 0 (5.28)
where τ is an external moment of force applied to the rod.
In this case, the resulting model can be composed from four pure transformers as shown in Fig. 5.20b
where
nxϕ = −ρ sin ϕ nyϕ = ρ cos ϕ nxρ = cos ϕ nyρ = sin ϕ
5.2.3 Transformations
An example of the magnitude of the error introduced by neglect of the motion of the earth may be cited
for the case of a particle that is allowed to fall from rest (relative to the earth) at a height h above
the ground. We can show that the rotation of the earth gives rise to an eastward acceleration (Coriolis
acceleration) relative to the earth and, neglecting air resistance, that the particle falls to the ground a
distance s
2 2h3
x= ω cos γ
3 g
east of the point on the ground directly under that from which it was dropped. The angular velocity of
the earth is ω = 0.729(10−4) rad/s, and the latitude, north or south, is γ. At a latitude of 45◦ and from
a height of 200 m, this eastward deflection would be x = 43.9 mm.
These corrections are negligible for most engineering problems which involve the motions of structures
and machines on the surface of the earth. In such cases, the accelerations measured with respect to
reference axes attached to the surface of the earth may be treated as ”absolute,” and Eq. 3/2 may be
applied with negligible error to experimental measurements made on the surface of the earth.* If the ideal
experiment described were performed on the surface of the earth and all measurements were made relative
to a reference system attached to the earth, the measured results would show a slight discrepancy upon
substitution into Eq. 3/2. This discrepancy would be due to the fact that the measured acceleration
would not be the correct absolute acceleration. The disc The discrepancy would disappear when we
introduced the corrections due to the acceleration components of the earth.
vxy 0 N Fxy
= · (5.34)
Fρϕ −Nt 0 vρϕ
where
vx vρ Fx Fρ
vxy = vρϕ = Fxy = Fρϕ = (5.35)
vy vϕ Fy Fϕ
and
cos ϕ − sin ϕ
N= (5.36)
sin ϕ cos ϕ
Electrical systems
dWBA dq dWBA
P= = = i vBA (6.3)
dt dt dq
6-1
6-2 MODULE 6. ELECTRICAL SYSTEMS
where i is the current flowing through the element and vBA is the voltage difference across it. In the
MKS system of units, power is measured in watts. One watt is equal to one joule/sec or one N.m/sec.
The electrical energy E supplied to an element is the time integral of the power
Z tb
E= i vBA dτ (6.4)
ta
where ta and tb are the beginning and end of the time interval during which power flows. If energy is
delivered to an electrical system through more than one pair of terminals, the total energy supplied is
the sum of the energies supplied at all the terminals.
The law of conservation of energy, or the first law of thermodynamics, states that the energy (of
all forms) which is delivered to a system must either be stored in the system or transferred out of the
system.
Figure 6.1: Symbols of pure source of electrical (a) voltage and (b) current.
Fig. 6.1a gives our graphical symbol representing a pure source of electrical voltage v(t) in multipole
diagrams. The two short line segments sticking out of the symbol are called pins. They denote poles of
the source representing terminals of the modelled real source of electrical energy. The symbol polarity
is indicated by the + sign. Besides the prescribed electrical voltage v(t), the pure source of voltage is
also associated with its current i(t). This current is imposed on the source by the system in which the
source operates. The assumed positive orientation of v(t) and i(t) in relation to the symbol polarity is
shown in Fig. 6.1a by the empty-head and full-head arrow, respectively.
The chosen orientation of source variables is such that the power of the source corresponds to the flow
of energy into the source. Sources are usually used to model supplies of energy into systems. Despite of
this, we shall always use the same ”consumer” variable-orientation convention for all the physical elements
regardless if they consume or deliver energy. Such a unified approach is for the benefit of modeling and
analysis simplicity. Note, that if the source symbol is oriented properly in a multipole diagram there is
no need to denote in it the variable orientation by arrows.
Figure 6.2: Symbol of ideal indicator of electrical (a) voltage and (b) current.
Pure sources of zero electrical voltage can be utilized as physical elements modeling ideal measuring
instruments or indicators of electrical current. Our graphical symbol for such an indicator is given in
Fig. 6.2b. Direction of the arrow showing the assumed positive polarity of the indicated electrical current
is fixed to the polarity of the symbol.
Note the difference between measuring electrical voltage and electrical current. The direct measure-
ment of electrical currents requires disconnecting the system and connecting it back by a electrical current
indicator. Electrical voltages are measured by indicators applied to systems without any need to dis-
connect them. This is the reason why electrical voltages represent electrical across variables while
electrical currents are electrical through variables.
Figure 6.3: (a) Real electrical resistor. Pure electrical (b) resistor and (c) conductor. (d) Model of real
resistor.
Fig. 6.3b gives our symbol for pure electrical resistors while in Fig. 6.3c is our symbol for pure
conductors. Polarity of both symbols is denoted by the + sign, and the assumed positive orientation
of the element voltage v and current i are associated with the symbol polarities. In Fig. 6.3d, the pure
conductor is used to model the real resistor shown in Fig. 6.3a. Note that we do not need to show in
the model the orientation of variables by arrows thanks to the unambiguous relation between the symbol
polarity and the variables.
Electric and magnetic field effects are assumed to be zero in pure resistors and conductors. These
elements can store neither electric-field nor magnetic-field energy. Characteristics of real components
6-4 MODULE 6. ELECTRICAL SYSTEMS
called conductors or resistors utilized in electrical and electronic systems, however, comply often quite
well with (6.7) in a fairly wide range of voltages and currents.
For example, if a voltage changing slowly enough is applied across a straight piece of a wire, the
electrical current passing through the wire can be approximated by (6.7), and the electrical conductance
of the wire
A
Ge = γ (6.8)
ℓ
where γ [S.m−1 ] is the specific conductivity of the material from which the wire is made, A is the
cross-section area of the wire, and ℓ is the wire length.
Electrical power is delivered to a resistance whenever current flows through it. For an ideal resistance
P = Re i2 = v 2 /Ge (6.9)
Power always flows into the resistor regardless of the signs of i and v. The power transferred to a
resistance cannot be retrieved and is said to be dissipated. Usually the power is transformed into thermal
internal energy in the resistance and the temperature rise which accompanies the increase in internal
energy results in flow of heat to the environment.
Figure 6.4: (a) Parallel-plate capacitor. (b) Pure electrical capacitor. (c) Model of real capacitor.
Most electrical components exhibit the property of capacitance to some extent. A device designed
to behave primarily like a capacitance is called an electrical condenser or capacitor. As an example,
Fig. 6.4a shows a real capacitor consisting of two parallel flat plates separated by a dielectric.
To permit this capacitive effect to be separated from other electrical phenomena, we shall define a
pure capacitor. It is a two-terminal electrical element in which charge q is a function of the voltage
difference across the element:
q = f (v)
In a pure capacitor, the conductors offer zero resistance to flow of charge, the dielectric separating the
conductors has infinite resistance to the flow of charge, and magnetic field effects are absent.
Fig. 6.4b shows the symbol to be used for a pure capacitor. The assumed positive direction of current
flow i and voltage drop v is indicated by an arrows drawn beside the symbol. This sign convention is
related to the polarity of the symbol indicated by the + sign. In Fig. 6.4c, the pure capacitor is used to
model the real capacitor. Note the relation between the orientation of the symbol and the variables in
the model.
The constitutive relationship for an ideal capacitor has charge proportional to voltage difference
q = C ·v
where C is the capacitance of the element, having units of farads [F] (F = A.s/V ). The elemental
equation for an ideal capacitor of constant C is therefore
dv
i(t) = C (6.10)
dt
6.5. PURE ELECTRICAL INDUCTOR 6-5
or
t
1
Z
v(t) = i(τ ) dτ + v(t0 ) (6.11)
C t0
Figure 6.5: (a) Electrical coil. (b) Pure electrical inductor. (c) Coil model.
An electrical element is said to be a pure electrical inductor provided that the magnetic flux Φ is
a function of the current i in the element generating the flux,
Φ = Φ(i) (6.12)
The function Φ(i) depends on the magnetic properties of the medium in the vicinity of the current flow.
By winding a wire in a geometry such as the helical coil of Fig. 6.5a, it is possible to reinforce the
flux inside the coil and to have the same flux link several turns of the coil. If a flux Φ links N turns, the
flux linkage λ is
λ = N Φ = f (i) (6.13)
The unit of flux linkage is weber [W] (1 W = 1 V.s).
As shown in Fig. 6.6, the flux linkage λ can be proportional to i, or it can be highly nonlinear
depending if the medium surrounding the inductor is, or is not, nonmagnetic.
The relationship of the magnetic flux linkage to terminal voltage of an inductor is given by Faraday’s
law
dλ(t)
v(t) = (6.14)
dt
6-6 MODULE 6. ELECTRICAL SYSTEMS
A voltage is ‘induced’ in the inductor which is equal to the rate of change of the flux linkages in the
inductor. Substituting (6.13) into (6.14), the terminal voltage can be written
dλ di
v(t) = · (6.15)
di dt
A linear pure electrical inductor, called an ideal electrical inductor, has flux linkage proportional
to its current:
λ(t) = Le · i(t) (6.16)
The electrical inductance Le of the ideal inductor is measured in henries [H] (1 H = 1 V.s/A), and
λ is defined to be zero when i = 0. As with capacitance, the computation of the inductance for a given
structure will usually require the use of field theory. It is noted that inductance is determined purely by
geometry and material properties.
For example, the inductance of a closely wound, long helical coil as that shown in Fig. 6.5a can be
estimated as
πN 2 µd2
Le =
4l
where l is the coil length, d is its diameter, N is the number of its turns, and µ is the permeability of the
material in the vicinity of the coil. For a vacuum (or very nearly for air) µ = µ0 = 4 × 10−7 H/m.
The inductance of a coil in air will be nearly linear, but the magnitude of the inductance is very
small. This magnitude may be increased by factors up to 106 by the use of ferromagnetic material such
as iron, cobalt, and nickel. However, inductors which use ferromagnetic materials are nonlinear and
give responses somewhat like the nonlinear curve shown in Fig. 6.6. The flattening of this curve as i is
increased is called saturation.
Inductance of a toroidal coil made of a conducting wire wound around a toroid made of a nonmetallic
material such as wood is approximately
N 2A
L e = µ0 [H] (6.17)
l
where N is the number of turns, A is the cross-sectional area, and l is the midcircumference along the
toroid.
The elemental equation for an ideal inductor of constant Le follows from (6.14) and (6.16):
d
v(t) = Le i (6.18)
dt
or
t
1
Z
i(t) = v(τ ) dτ + i(t0 ) (6.19)
Le t0
An inductor stores energy in the magnetic field associated with the current. The electrical energy
stored in a pure inductor is called the magnetic-field energy
Z tb Z λb
Em = v idτ = idλ (6.20)
ta λa
Figure 6.7: (a) Pair of interacting coils, (b) coupled electrical inductors.
For the following discussions, it will be more convenient to consider magnetic linkages of the coils
instead of their fluxes. Recall, that if all the N1 turns of the first coil are linked by the flux φ1 , the flux
linkage of the first coil is λ1 = N1 φ1 . Similarly, if all the N2 turns of the second coil are linked by the
flux φ2 , than λ2 = N2 φ2 .
Assuming further, that the magnetic fluxes are linear functions of the coil currents, we can express
the magnetic linkages of the two coils as
λ1 (t) = L11 i1 (t) + L12 i2 (t)
λ2 (t) = L21 i1 (t) + L22 i2 (t) (6.23)
where L11 and L22 , is the self-inductance of the first and second inductor, respectively. L12 = L21 are
the mutual inductances of the inductors. Thus the two coils can be modelled by a pair of coupled linear
electrical inductors shown in Fig. 6.7b. Their constitutive relation (6.23) rewritten in matrix form is
λ1 (t) i1 (t) L11 L12
=L· where L = (6.24)
λ2 (t) i2 (t) L21 L22
6-8 MODULE 6. ELECTRICAL SYSTEMS
Figure 6.8: (a) Three interacting coils, (b) coupled electrical inductors.
Figure 6.9: Models equivalent to coupled electrical inductors employing an ideal transformer.
In the model given in Fig. 6.9b, the ideal-transformer ratio and the inductances are
L12 M2 M2
n= Lm = La = L11 − (6.32)
L22 L22 L22
Lm is called the magnetizing inductance as it models the effect of the magnetic flux common to both
coils. La is the leakage inductance modeling the effect of the leakage fluxes seen at the first port. The
equivalence of both models to a pair of coupled inductors can be easily verified by their analysis.
It is interesting to investigate the model in Fig. 6.9b for the limit case when L11 → ∞, L22 → ∞,
and k → 1. From (6.32) we can see that under these assumptions La → 0 and Lm → ∞. Hence, in the
limit case, the inductance La becomes a closed circuit while the inductance Lm becomes an open circuit.
Also, in the limit case, n = ±N2 /N1 assuming that N1 and N2 are the coil turns fully linked by their
fluxes. The sign of n can be determined in the same way as the sign of the mutual inductance.
6-10 MODULE 6. ELECTRICAL SYSTEMS
Module 7
Fluid systems
Figure 7.1: Pressure across (a) a flow-through component, (b) an open tank supplied by liquid.
Fig. 7.1a shows a fluid-system component with two inlets. The bulk values of the gauge pressures at
the inlets A and B are denoted as pA and pB , respectively. The empty-head arrow in Fig. 7.1 indicates the
assumed positive polarity of the difference pBA = −pAB between the inlet gauge pressures. The pressure
difference
pBA = pB − pA (7.1)
is positive if pB > pA . Such a pressure difference is called the pressure drop across the component.
Another fluid-system component with two inlets is shown in Fig. 7.1b. It is an open tank in the form
of a cylinder of the cross sectional area A filled with a liquid coming to the tank via the inlet C. The
liquid is leaving the tank through the opening D situated in the tank wall close to its bottom. Recollect
that the pressure at any point in a fluid at rest is the same in all directions. If the fluid flow through the
7-1
7-2 MODULE 7. FLUID SYSTEMS
opening D is slow enough and the diameter of the opening is negligible when compared with the hight h
of the liquid in the tank above the tank bottom, we can assume that the pressure pD at the opening D
is the same as the pressure acting on the bottom of the tank. Thus,
mg
pD = (7.2)
A
m = ρV (7.3)
where ρ is the mass density of the fluid defined as the mass of the fluid per unit volume with units
kg/m3 .
Therefore,
V ρg
pD = = hρg (7.4)
A
as the volume of the liquid in the tank is V = A h.
dm
ṁ = = ρ vA
dt
where A [m2 ] is the cross-sectional area of the pipe or duct normal to the flow direction, ρ is the average
mass density over A, and v is the average value of the fluid velocity in the flow direction over A.
Figure 7.2: Inlet volume flow of two-inlet components: (a) flow-through component, (b) open tank
supplied by liquid.
As measuring of volume-flow rate is much easier then measuring mass-flow rate, engineers prefer the
former variable when analyzing fluid systems. Volume-flow rate Q is the variable which expresses the
volume of fluid passing a given area per unit time, so its unit is m3 /s. This quantity, denoted in Fig. 7.2
by full-head arrows, is given as
dV
Q(t) = = vA (7.5)
dt
where V is the fluid volume and v is the bulk velocity of the fluid flow through the area. Note, that the
relation between the mass and volume flow rates is
ṁ = ρ Q
7.2. MODELING OF FLUID SYSTEMS 7-3
dW = p A dx
The quantity of volume dV of fluid passing A is
dV = A dx
Therefore
p = dW/dV (7.6)
Thus pressure is the work done in passing a unit volume of fluid across the area on which p acts.
Power, as we know, is the rate of flow of work. Hence from (7.5) and (7.6), the fluid power
dW dV
P= =p = pQ
dt dt
The net power delivered to a two-inlet component shown in Fig. 7.1 is
P = pA Q A + pB Q B
If the fluid and the component is such that QA + QB = 0, i.e. if the flow in equals to flow out,
P = pAB QA = pBA QB
The fluid energy delivered to a fluid-system component is the time integral of the power.
The line segment shown in Fig. 7.3b represents a segment of an ideal pipe with inlets A and B
containing an ideal fluid such that
pA = pB and QA + QB = 0
Symbols of fluid-system components with their pins interconnected by ideal fluid links form a multipole
diagram of the system. The multipole diagram in Fig. 7.4a shows the way of measuring pressures of a two-
inlet component using ideal pressure-measuring instruments. The way of direct measuring of volume-flow
rate between elements is shown in Fig. 7.4b.
Note, that pressure is the across-variable in fluid systems. To measure the volume-flow rate between
components, we must sever the pipe interconnecting the components at the section of interest and include
7-4 MODULE 7. FLUID SYSTEMS
Figure 7.4: (a) Measurement of (a) element pressures and (b) of volume-flow rate between elements.
a flow-rate measuring instrument to renew the connection. It is thus evident that the volume-flow rate
is the through-variable.
Relation (7.1) is an example of the geometric constraints which are placed on pressures in a system
due to continuity of space. It is the expression of the postulate of pressure compatibility – called
also the axiom of pressure composition – for the two-inlet subsystem.
Mass, like energy, is a conserved property, and it cannot be created or destroyed. Mass m and energy
E can be converted to each other according to the famous formula proposed by Einstein: E = m c2 where
c = 2.998.108 m/s is the speed of light. This equation suggests that the mass of a subsystem may change
when its energy changes. However, for all energy interactions encountered in engineering practice, with
the exception of nuclear reactions, the change in mass is extremely small and cannot be detected by even
the most sensitive devices.
We can thus assume validity of the principle of conservation of mass according to which the mass
entering a region in space minus the mass leaving it must equal the change of mass stored within the
region.
Figure 7.5: Threepole model of a two-inlet component to satisfy the postulate of continuity.
Let us consider the component with two fluid inlets shown in Fig. 7.2. The volume flow rates at the
component inlets A and B are denoted as QA and QB . The assumed positive directions of the fluid flows
are indicated by the full-head arrows.
The principle of conservation of mass for the two-inlet component
where ṁA = ρ QA and ṁB = ρ QB is the mass-flow rate into the inlet A and B, respectively. As m = ρV
is the mass of the fluid contained in the component if V is the fluid volume of the component,
where ρ̇ = d ρ/dt and V̇ = dV /dt is the time rate of the mass density and component volume, respectively.
Unlike mass, fluid volume is not a conserved property as (7.8) indicates. Only if ρ̇ = 0, i.e., if the fluid
is incompressible, and if V̇ = 0, i.e. if the fluid volume in the component is constant, the flow entering
the component through one inlet equals the flow leaving the component through the other inlet
QA + QB = 0, or QA = −QB (7.9)
This is the postulate of continuity expressed for the fluid flow into the two-inlet component, which
is valid under the above assumptions only. If the fluid cannot be considered as incompressible, or the
container volume as constant, we must respect accumulation of fluid mass in the component by adding
a third pole C to the two-inlet component model linked to the reference as shown in Fig. 7.5. To satisfy
the postulate of continuity, the volume-flow rate of the third pole must be
QC = −QA − QB
7.3. FLUID ACCUMULATION 7-5
Fig. 7.6b shows three ideal pipe segments passing fluid freely between any two of the inlets A, B and
C via the interconnection D. Such an ideal-pipe structure is characterized by the relations
pA = pB = pC QA + QB + QC = 0
where the latter statement is the expression of the postulate of continuity for the interconnection C.
The relationship between the volume-flow rate of fluid into the capacitor and the fluid pressure at the
capacitor inlet is
dp
Q(t) = Cf · (7.12)
dt
or Z t
1
p(t) = Q dτ + p(t0 ) (7.13)
Cf t0
where p(t0 ) is the capacitor pressure at t = t0 .
Figure 7.7a gives our symbol for pure fluid capacitors. Only one the pole representing the fluid intake
of the capacitor is shown in the symbol. This pole is associated with the pressure p at the intake and
with the volume-flow rate Q(t) into the capacitor. The second pole is hidden as it must be always fixed
to the pressure reference.
The symbol consists of a square representing the capacitor volume and pole representing the capacitor
inlet. The capacitor is associated with two variables, the capacitor pressure p measured with respect to
a constant pressure reference, and the capacitor volume-flow rate Q. The assumed positive polarity of
p is indicated by the empty-head arrow placed between the symbol for the pressure reference and the
7-6 MODULE 7. FLUID SYSTEMS
symbol pole. The orientation of the full-head arrow shows that the positive Q is always directed into the
capacitor.
In general, a fluid capacitor can be nonlinear and time-variable. Then it is characterized by the
relation
dV ∂V dp ∂V
Q(t) = = · + (7.14)
dt ∂p dt ∂t
Figure 7.7b shows V -p characteristics of a nonlinear and linear fluid capacitors specified by their
constitutive relations (7.10) and (7.11).
Power P supplied to a capacitor is
P = Qp (7.15)
providing that the fluid density is constant.
Energy Ep , called fluid potential energy and stored in a fluid capacitor by virtue of the pressure, is
Z t
Ep = P dt (7.16)
0
dV = Cf dp (7.17)
p
Cf p2 V2
Z
Ep = Cf p dp = = (7.18)
0 2 2Cf
Figure 7.8: (a) Open cylindrical tank, (b) its fluid capacitor model.
The fluid pressure by the liquid level in the open tank is zero as it is the ambient reference pressure.
If there are no fluid accelerations in the tank except due to gravity, the fluid pressure by the tank bottom
equals to the weight of the liquid per the tank horizontal area, or
mg
p= = ρgh (7.20)
A
where m = ρ A h [kg] is the liquid mass, ρ [kg/m3 ] is the liquid mass density, and g [m/s2 ] is the
acceleration due to gravity.
Substituting p for h from (7.20) and comparing the result with (7.11) we obtain the open tank
capacitance as
A
Cf = (7.21)
ρg
Fig. 7.8b shows an ideal fluid capacitor of the fluid capacitance Cf modeling dynamics of the open
tank. In Fig. 7.9a, the liquid volume flow Q is supplied into the open tank by a pump. This situation
is modelled in Fig. 7.9b where the pump is represented by an independent source of fluid flow. The
relationship (7.20) between the height h of the liquid level and the pressure pA in the tank is expressed
in the model using a scaling block.
7.3. FLUID ACCUMULATION 7-7
Figure 7.9: (a) Open tank supplied by a pump, (b) model of the fluid system.
Figure 7.10: (a) Open spherical tank, (b) its V -p characteristic and (c) capacitance vs. liquid-level height.
Figure 7.11: (a) Tank containing compressed fluid. (b) Model of the pressurized tank.
where ρ [kg/m3 ] is the fluid density and V0 [m3 ] is the constant volume of the tank.
To employ tank pressure p rather than the mass density ρ as a variable, we must utilize the equation
of state of the fluid. For most liquids, a reasonably good first-order approximation for their equation of
state is
ρ
dρ = dp (7.26)
β
where β [N/m2 ] is the bulk modulus of elasticity of the liquid.
For most gases including air,
ρ
dρ = dp (7.27)
np
where p is the absolute fluid pressure and n is a constant whose value is between 1.0 and 1.4, depending on
how rapidly the fluid is compressed. If the pressure variations in the gas are small, it may be considered
to have a bulk modulus of elasticity β ≈ n p.
Substituting (7.26) into (7.25) yields
ρ dp
QA = (7.28)
β dt
Comparing this result with (7.14) indicates that the pressurized tank behaves like an ideal fluid capacitor
with the capacitance
V0
Cf = (7.29)
β
Under the above assumptions the pressurized chamber given in Fig. 7.11a can be represented graphi-
cally as shown in Fig. 7.11b. The capacitor pole representing the accumulator inlet is coalesced with node
A of a system model. The capacitor pressure p corresponds to the gauge pressure pA of the accumulator.
where ρ, p, and ϑ are the mass density, pressure, and temperature, respectively, of the liquid about initial
values of ρ0 , p0 , and ϑ0 . A more convenient form of (7.33) is
1
ρ = ρ0 1 + (p − p0 ) − α(ϑ − ϑ0 ) (7.34)
β
where
∂p 1 ∂ρ
β ≡ ρ0 and α ≡ −
∂ρ ϑ ρ0 ∂ϑ p
Equation (7.34) is the linearized equation of state for a liquid. The mass density increases as pressure
is increased and decreases with temperature increase. Because mass density is mass divided by volume,
equivalent expressions for β and α are
∂p 1 ∂ρ
β = −V0 and α =
∂ρ ϑ V0 ∂ϑ p
where V is the total volume and V0 is the initial total volume of the liquid.
The quantity β is the change in pressure divided by the fractional change in volume at a constant
temperature and is called the isothermal bulk modulus or simply bulk modulus of the liquid. It is
the most important fluid property in determining the dynamic performance of hydraulic systems related
to the ‘stiffness’ of the liquid. Note that it is always a positive quantity. The reciprocal of β is often
termed the compressibility of the liquid.
An adiabatic bulk modulus, βa may also be defined. It can be shown that the adiabatic bulk
modulus is related to the isothermal bulk modulus by
Cp
βa = β (7.35)
Cv
where Cp /Cv is the ratio of specific heats. Because this ratio is only slightly in excess of unity for a
liquid, the distinction between adiabatic and isothermal moduli is insignificant.
The quantity α is the fractional change in volume due to a change in temperature and is called the
cubical expansion coefficient of the fluid.
Q = f (p) (7.36)
P = p Q = Rf Q2 = p2 /Rf
Figure 7.12: (a) Pipe segment. (b) Pure fluid resistor and (c) conductor. (d) Pipe model.
Fig. 7.12b gives our graphical symbol for pure fluid resistors. The pure conductor is associated with
two variables: the conductor pressure drop p and the conductor fluid flow Q. Polarities of these two
variables are always related to the polarity of the symbol indicated by the + sign as shown in Fig. 7.12b.
This means that the pressure at the + pole is assumed larger than that at the second pole. The conductor
flow is considered positive if it is oriented from the + pole towards the conductor.
The symbol of the pure fluid conductor is utilized in Fig. 7.12c to model the pipe segment shown
in fig Fig. 7.12a. The nodes A and B represent energy interactions at the corresponding inlets of the
real pipe segment. The gauge pressures pA and pB at the pipe inlets are denoted by empty-head arrows
placed between the symbols for the reference pressure and the related nodes. Respecting the conductor
orientation, the pressure drop pBA corresponds to the conductor pressure p, and the flow QB corresponds
to the conductor flow Q. As the postulate of continuity applies to the pure conductor QA + QB = 0, or
QA = −QB .
where ν is the absolute viscosity of the fluid, ℓ is the length and d is the inside diameter of the capillary.
The ideal fluid resistor is applicable if the tube is long enough in relation to its diameter (ℓ ≫ 100d)
so that the effects of nonuniform flow near the tube entrance are negligible. At the same time, the rate
of change of volume flow must be small enough that inertia effects are negligible.
Also the effect of an incompressible flow through a porous plug can be approximated by an ideal
fluid conductor. In this case the constitutive relation (7.37) is expressed by the D’Arcy’s law. The
corresponding linear resistance Rf is usually determined for a given porous medium and working fluid
experimentally.
Example
Fig. 7.13a shows a water tank which discharges to the atmosphere through a porous plug. Recollect
that such a fluid restrictor produces a pressure drop linearly proportional to the flow through it. A
dynamic model of the tank is shown in Fig. 7.13b. Node A represents the interconnection of the plug
with the tank. Note that the direction of the volume-flow rate Q is in agreement with the direction with
the flow of the conductor but opposite to that of the capacitor.
The turbulent character of the flow is ensured only at large enough Reynolds numbers. At smaller
values of Re the flow becomes laminar and the discharge coefficient varies with Re and depends on the
geometry of the orifice edges. It is common in practice, however, to use the turbulent flow equation (7.43)
regardless of the Reynolds number unless flows through very small openings are studied.
or
d
p(t) = Lf Q(t) (7.45)
dt
where r(t) is the inductor pressure momentum, p(t) is the inductor pressure, Q(t) is the inductor volume
flow, and the constant parameter Lf [Pa.s2 m−3 ] is the fluid inductance of the fluid inductor, also termed
fluid inertance.
Figure 7.14: (a) Pipe segment. (b) Pure fluid inductor. (c) Pipe model.
To visualize the concept of fluid inductance, let us consider the unsteady frictionless flow of an
incompressible fluid in a nonaccelerating pipe segment of length ℓ. If the pipe has constant area cross-
section A and the velocity v of the fluid is uniform across any cross section of the pipe, we can say that
every fluid particle has the same velocity v and hence the same acceleration dv/dt. Then the force F
necessary to produce an acceleration dv/dt of the fluid mass in the pipe is
dv
F (t) = ApAB (t) = Aℓρ
dt
where pAB (t) is the pressure drop across the pipe segment. As ρ [kg.m−3 ] denotes the mass density of
the fluid, ρAℓ is the mass of all the fluid in the pipe segment. The volume flow QA (t) = Av(t), so that
the fluid inductance of the flow in the pipe is
ℓ
Lf = ρ (7.46)
A
This relation holds only when the pipe is not being accelerated. If the pipe itself has an acceleration,
additional pressure difference effect between the inlets of the pipe segment will have to be taken into
account.
Fig. 7.14b gives our graphical symbol for the pure fluid inductor (or resistor). The pure inductor is
associated with two variables: the inductor pressure drop p and the conductor fluid flow Q. Polarities of
these two variables are always related to the polarity of the symbol indicated by the + sign as indicated
in Fig. 7.14b. This means that the pressure at the + pole is assumed larger than that at the second pole.
The inductor flow is considered positive if it is oriented from the + pole towards the inductor.
The symbol of the pure fluid inductor is utilized in Fig. 7.14c to model the pipe segment shown in
fig Fig. 7.14a. The nodes A and B represent energy interactions at the corresponding inlets of the real
pipe segment. The gauge pressures pA and pB at the pipe inlets are denoted by empty-head arrows
placed between the symbols for the reference pressure and the related nodes. Respecting the conductor
orientation, the pressure drop pAB corresponds to the conductor pressure p, and the flow QA corresponds
to the conductor flow Q. As the postulate of continuity applies to the pure inductor QA + QB = 0, or
QA = −QB .
7.6. FLUID TRANSDUCERS 7-13
In actual fluid piping, significant friction effects are often present along with the inertance effects, and
the inertance effect tends to predominate only when the rate of change of flow rate (fluid acceleration) is
relatively large. Since flow resistance in a pipe decreases more rapidly with increasing pipe area A than
does inertance, it is easier for inertance effects to overshadow resistance effects in pipes of large sizes.
However, when the rate of change of flow rate is large enough, significant inertance effects are sometimes
observed even in fine capillary tubes.
A fluid-flow system consisting of two interconnected tanks is shown in Fig. 7.15a. The corresponding
model respecting fluid-flow inertia is in Fig. 7.15b.
pA AA = pB AB
where pA and pB are presures exerted on the piston by the fluid in each of the chambers. To obtain a
constitutive relation characterizing the pure model of the device shown in Fig. 7.16b we must consider
this expression together with the energy-transfer relation taking on the form
pA Q A + pB Q B = 0
where QA and QB are the fluid volume flows into the chambers through the inlets A and B. Thus the
constitutive relation is
pA (t) 0 n QA (t)
= ·
QB (t) −n 0 pB (t)
ẋA FA + ẋB FB = 0
we obtain the constitutive relation of the rectilinear-motion transformer pure model shown in Fig. 7.17b
ẋA (t) 0 n FA (t)
= ·
FB (t) −n 0 ẋb (t)
where
ẋA FB Ab
n= =− =
ẋB FA Aa
Figure 7.18: Three-land-three-way spool valves with cylinder load (S ... supply, R ... return).
7.6. FLUID TRANSDUCERS 7-15
Thermal systems
8-1
8-2 MODULE 8. THERMAL SYSTEMS
H = f (ϑ) (8.4)
H = Ct ϑ (8.5)
Ct = cp m (8.6)
where m [kg] is the body mass, cp [J/kg] is the specific heat of the body material.
The symbol for a pure thermal capacitor is shown in Fig. 8.1a . As for a fluid capacitor and inertor,
one pole of the thermal capacitor must be always associated with a reference temperature which is zero
or constant. As no heat occurs through the reference “connection” the reference is not shown in the
symbol.
For a body to be modelled by an ideal thermal capacitor, all thermal resistance effects (discussed
below) should be negligible within it (usually in comparison with external resistances) so that the tem-
perature will be uniform throughout the mass which comprises the capacitor. A mass of metal or a
well-stirred tank of fluid often behave very nearly like ideal thermal capacitors.
8.4. HEAT TRANSFER 8-3
Figure 8.1: Symbol of pure thermal (a) capacitor, (b) resistor and (c) conductor. (d) Symbol of thermal
reference.
ϑ = f (Ḣ) (8.8)
ϑ = Rt Ḣ (8.9)
Ḣ = Gt ϑ (8.10)
A given material may be considered as a pure or ideal thermal resistance only if the amount of heat
stored in it is negligible or does not change appreciably during operation. Fig. 8.1b shows the symbol for
pure thermal resistances.
When two sheets of solid material are placed together with imperfect contact (which is typical), the
total resistance of the combination is greater than the sum of the individual resistances of each of the
sheets because of the small air space (or vacuum or oxide film in some cases) which exists at the interface.
Although the thickness of the air space is very small, this extra resistance can be appreciable, because
still air is a relatively poor conductor of heat.
The notion of an over-all coefficient of heat transfer for a given physical arrangement of components
in a thermal system is frequently employed, and it is defined as
Ḣ
ch = (8.13)
Aϑ
where ch is the coefficient of heat transfer [kg.cal/oC.s.m2 ], A is the cross-sectional area, and ϑ is the
over-all temperature drop.
A determination of the heat-transfer coefficient for a given thermal situation is usually very complex
and nearly always involves experimental measurements. The coefficient ch is not constant for a given
geometry, but may often be assumed constant over limited ranges.
Equation (8.13) describes an ideal thermal resistance Rt whose value is
1
Rt = (8.14)
ch A
where cr is a constant which depends on the shape, area, and surface characteristics of the body which is
emanating or receiving radiation, and on the wavelength of the radiation. The temperatures ϑA and ϑB
must be the absolute temperatures in K of the surfaces emitting and receiving radiation, respectively.
Fig. 8.4, illustrates schematically radiative heat transfer. Here it is assumed that the radiant energy
emitted by surface A is absorbed by surface B.
The constant cr in (8.15), measured in W/K4 is very small, so that heat transfer by radiation is usually
insignificant unless large temperatures are involved. As an example, for radiation from a spherical black
body to a black environment which completely surrounds it, the value of cr is given by σAo , where σ is
the Stefan-Boltzmann constant, σ = 1.37 × 10−5 [W.m−2 K−4 ] and Ao is the area of the radiating sphere
in m2 .
The expression (8.15) for heat transfer is nonlinear and does not represent a pure thermal resistance
as we have defined it, because Ḣ does not depend on temperature difference. If the temperatures ϑA and
ϑB change by only small amounts from mean values ϑA0 and ϑB0 , (8.15) may be linearized.
Let Ḣ0 be the heat flow when ϑA = ϑA0 and ϑB = ϑB0 , let ϑB be constant and ∆ϑA be a small
change (compared with ϑA0 and ϑB0 ) of temperature ϑA . Then ∆Ḣ is the change in heat flow
difference is applied. This effect has been used for temperature measurement and for the generation of
small amounts of electric power. When energy flows from the electrical system to the thermal system, re-
frigeration or cooling can be obtained. Small thermoelectric refrigerators have been made for the cooling
of electric components and for use as automatically controlled constant-temperature references.
Module 9
Magnetic systems
If the core is disrupted by an air gap, as in Fig. 9.1b, the flux lines indicate that large leakage and
fringing fluxes result and the flux within the core varies with the distance from the coil. Still, such a coil
can be modeled as a magnetic circuit if the largest part of the flux lines is concentrated into the core
which determines their geometric shape. In the contrary, the short cylindrical coil shown in Fig. 9.1c
cannot be modeled as a magnetic circuit. Except for the short part of the flux lines within the core, most
of their length is situated in the air.
Magnetic voltage (called also magnetomotive force) w(t) [A] plays the role of the across variable in
magnetic interactions. The concept of magnetic voltage is applicable in quasi-magnetic fields when the
rate of change of electric field normal to the subsystem energy boundary is negligible.
Magnetic flux rate ψ̇(t) [Wb/s] into a subsystem through an energy entry is defined as
dφ
φ̇(t) =
dt
where φ(t) is the total of magnetic flux B [Wb/m2 ] over the entry area passing through it in the perpen-
dicular direction to it.
Magnetic interactions are also governed by laws similar to those of Kirchhoff. The continuity postulate
applies to the rate of magnetic flux into the entries of a subsystem assuming the magnetic flux leakage
through the subsystem boundary is negligible. If it is not so, the boundary must be enlarged to include
also the leakage flux lines.
9-1
9-2 MODULE 9. MAGNETIC SYSTEMS
Fig. 9.2a shows branching of the magnetic flux in a ferromagnetic material. The dynamic diagram
representation of the branching is in Fig. 9.2b. The interconnection portrayed there by node A with three
line segments representing ideal magnetic conductors.
Obviously, the magnetic flux in magnetic systems obeys the continuity postulate based on the law of
conservation of magnetic field. In general, if there are n magnetic conductors interconnected at a point,
and if φk is a magnetic flux of the k-th conductor, then
n
X
φk = 0 (9.1)
k=1
where the fluxes oriented towards the interconnection are considered as positive.
According to Ampere’s law the integral around any closed path ℓ in the magnetic field of intensity H
equals the sum of electrical current I passing through the path ℓ, i.e.
I
H · dℓ = I (9.2)
ℓ
Positive current is defined here as flowing in the direction of the advance of a right-handed screw turned
in the direction in which the closed path is traversed by H.
Assuming that the closed path ℓ is divided into n segments ℓk , k = 1, 2, . . . , n, then
I Xn Z
H · dℓ = Hℓ dℓ (9.3)
ℓ k=1 ℓk
where Z
Hℓ dℓ = wk (9.4)
ℓk
where wm = −I represents the total magneto-motoric voltage of ℓ. We may thus reformulate Ampere’s
law as stating that “the sum of magnetic voltage drops around any closed path is equal to zero,” which
represents the compatibility postulate for magnetic systems.
f (φ, w, t) = 0 (9.6)
where Cm , the magnetic capacitance, is known as the permeance, and is measured in henrys. Its reciprocal
value Dm = 1/Cm is called the reluctance.
9.2. PURE MAGNETIC CAPACITOR 9-3
where Bn is the magnetic flux density vector which is normal to the surface. Within a material, B and
H are related by the permeability µ of the material:
B = µH µ = µ0 µr (9.10)
where µ0 = 4π × 10−7 [H.m−1 ] is the permeability of free space, and µr [ – ] is the relative permeability
of the material. Permeability can be qualified as a scalar only in regions of homogenous and isotropic
materials. In case of nonmagnetic materials µr can be considered as equal to 1.0 for most of the practical
purposes. Properties of most of magnetic materials are strongly nonlinear, the expression (9.10) can then
be applied to small variations of H in the vicinity of a certain fixed operating point.
Figure 9.3: (a) Ferromagnetic core. (b) Symbol of pure magnetic capacitor. (c) Dynamic model.
Fig. 9.3a shows a piece of a core made of ferromagnetic material of a cross section area A and of
length ℓ. If we assume that the magnetic field intensity H is nonvarying along the length ℓ, and that the
flux density B is uniform over the area A, the permeance of the component becomes
φ(t) BA µA
Cm = = = (9.11)
w(t) Hℓ ℓ
which is similar to the expression for electrical capacitance in a region with similarly uniform electrical
properties.
Fig. 9.4a shows a torus made of highly permeable material with a winding of N turns. The magnetic
voltage around the magnetic path is w = N i. This magnetic voltage establishes a magnetic field intensity
H, which in turn produces a magnetic flux density B = µ H in the material. The narrow air gap
interrupting the closed magnetic circuit makes an example of a magnetic system components that can be
modeled as an ideal magnetic capacitor as shown in Fig. 9.4b.
Magnetic circuit with leakage fluxes forms a distributed-parameter system. As in the case of other
similar systems, it can be approximated by decomposing it into sections each of which is represented
by a lumped parameter model. The larger is the number of the sections, the more accurate is the
approximation. For example, the magnetic circuit in Fig. 9.5a is divided into a number of parts and
assumed fluxes both in the core and in the air are indicated. Fig. 9.5b shows the corresponding lumped-
parameter model in which the permeances are related to the core segments, to the leakages and to the
gap.
9-4 MODULE 9. MAGNETIC SYSTEMS
Figure 9.5: Magnetic circuit with distributed parameters and its models.
f (φ̇, w, t) = 0 (9.12)
where Gm , measured in ohms, is the magnetic conductance of the conductor. The ideal magnetic resistor
is an element characterized by the inverse relation
where Rm is the magnetic resistance measured in siemens. Such elements can be used for modeling eddy
current losses in magnetic systems, for example. Ideal magnetic conductors or resistors can be used to
model energy dissipation losses associated with small variations of magnetic field in some materials.
Note, that for the sake of consistency in the unified approach to modeling of dynamic effects in different
energy domains, we do not adopt the traditional definitions of magnetic conductance and resistance. That
is, we do not consider magnetic conductance to be equivalent to permeance, and magnetic resistance to
reluctance as it is common on most of the other sources.
Figure 9.6: (a) Electromagnetic coil, (b) magnetic model, (c) simplified magnetic model.
reluctances Df 1 , Df 2 and Df 3 represent this accumulation in the individual branches of the ferromagnetic
core. In Fig. 9.6c, the same model is simplified taking advantage of the core symmetry.
Analyzing the model in Fig. 9.6c we shall obtain
Ni
φ= (9.15)
Dm
Figure 9.7: (a), (b) Coil electromagnetic models, (c), (d) coil electrical models.
Fig. 9.7b differs from Fig. 9.7a by elements modeling some parasitic effects in the coil. The electrical
resistor R corresponds to the electrical resistance of the coil winding, the magnetic resistors Rf 1 and
9-6 MODULE 9. MAGNETIC SYSTEMS
Rf 2 stand for the energy losses in the segments of the ferromagnetic core, and the electrical capacitor C0
represents accumulation of electric energy within the coil winding.
Note, that the model in Fig. 9.7b cannot be described just by a set of algebraic equations as it was in
the case of the other coil models discussed so far – here we need a set of differential equations.
Analyzing the model shown in Fig. 9.7c we can arrive at the coil electrical model given in Fig. 9.7c,
i.e. in a pure inductor. According to the static definition of the coil inductance
λ
L= (9.16)
i
where λ = N φ is the flux linkage measured in webers. Taking into consideration (9.15), the coil inductance
can be expressed as
N2
L= (9.17)
Dm
Similarly, analyzing the model from Fig. 9.7b we can obtain parameters of the coil electrical model
with parasitic effect elements in Fig. 9.7d.
The small increase that occurs beyond this condition is due to the increase of the flux density in the
vicinity of the space occupied by the material. It is often convenient to subtract out this component of
‘free space’ flux density and observe only the flux density variation within the material. Such a curve
is known as the intrinsic magnetization characteristic and is of use in the design of permanent magnet
devices.
A symmetrical hysteresis loop is obtained after a magnetic field intensity reversal between plus and
minus Hs . The area within the hysteresis loop corresponds to the hysteresis losses, i.e. to the energy
dissipated in the material during one reversal cycle. This area varies with temperature of the material
and with the frequency of the reversal.
Eddy current losses are caused by electrical currents induced in the core, called eddies, since they
tend to flow in closed path within the magnetic material itself. The eddy current loss in a sinusoidally
9.7. PERMANENT MAGNET EXCITED SYSTEMS 9-7
Pe = ke f 2 Bm
2
[W] (9.18)
where Bm is the maximum value of flux density, f the frequency, and ke is a proportionality constant
depending upon the type of material and the lamination thickness.
To reduce eddy current loss, the magnetic material is laminated, that is, divided up into thin sheets
with a very thin layer of electrical insulation between each sheet. The sheets must be oriented in a
direction parallel with the flow of magnetic flux. It should be recognized that laminating a magnetic
material generally increases its volume. The ratio of the volume actually occupied by magnetic material
to total volume of a magnetic part is known as the stacking factor. This factor is important in accurately
calculating flux densities in magnetic parts.
Therefore, instead of Eq. (9.10), for larger variations of magnetic field intensity H the relation
B = f (H) (9.19)
Electromechanical systems
v = B ℓ ż (10.2)
The polarity of v is given by Fleming’s right-hand rule. That is, if conductor moves in the direction of
the force F , the polarity of v with respect to i is such that an electrical power is generated.
Figure 10.1: (a) Current-carrying conductor moving in a magnetic field, (b) its ideal-transformer model.
Equations (10.1) and (10.2) form together the constitutive relationship of an ideal energy-transfer
transformer which can be used for modeling this simple device as illustrated in Fig. 10.1b. The transformer
poles A and B represent electrical terminals of the wire, whereas pole Cz represents motion of the wire
along the z axis. Coupling ratio of the transformer is obviously
ż 1 1
n= = = (10.3)
v Bℓ B2πrN
10-1
10-2 MODULE 10. ELECTROMECHANICAL SYSTEMS
into the moving coil. The external force F acting on the coil is required to hold the coil in equilibrium
against the magnetic forces. The voltage across the voice coil is v.
The constitutive relations for the moving coil can be obtained from (10.1) and (10.2) by substituting
ℓ = 2π r N , where r is the coil radius and N is the number of turns of the coil winding. Thus the moving
coil can be modeled by the ideal energy-transfer transformer shown in Fig. 10.2b where the transforming
ratio
v
n = = 2π N B i (10.4)
ẋ
In the model shown in Fig. 10.2b we neglect the electrical resistance and self-inductance of the voice-coil,
as well as any capacitance. We also neglect the mechanical mass and friction, as well as any restraint
from the lead-in wires. These effects can be respected by augmenting the ideal transformer with physical
elements representing these effects.
ẏ = r ϕ̇ cos ϕ (10.5)
where r is the radius of the armature and ϕ̇ is its angular velocity. It is this component which must be
used in (10.2) for the voltage equation
v = B ℓ r ϕ̇ cos ϕ (10.6)
where ℓ is the effective length of all conductors cutting the flux lines with an average density value B.
Wires may be wound many times around the armature, and the total voltage generated is the sum of
all the voltages generated by each wrap. Note that voltage induced on one side of the rotor adds to that
on the other because although the direction of flux cutting is opposite on two ends of the rotor diameter,
the direction of current flow is also opposite on the two ends. The electric circuit can be completed with
clip rings as shown.
If current starts to flow through the rotor winding, a force given by (10.1) will act on it giving rise
to a torque opposing the rotation. The magnitude of the external torque required to hold the rotor in
equilibrium will be
τ = F r cos ϕ = −B ℓ r i cos ϕ (10.7)
This together indicates, that the alternator can be modeled by a pure transformer with the variable
ratio
ϕ̇ 1
n= = (10.8)
v B ℓ sin ϕ
as shown in Fig. 10.3b. Pole Cϕ represents the rotational motion of the rotor.
Real alternators have more complicated configurations than the one shown schematically in Fig. 10.3a
an may have sets of windings to generate three-phase power, but the basic modeling idea based on and
energy-transfer transformer still applies.
10.4. PERMANENT-MAGNET DC MACHINE 10-3
which can be represented, as in Fig. 10.4b, by a simple ideal-energy transfer transformer with the trans-
forming ratio
ϕ̇ 1
n= = (10.10)
v Bℓr
There is some ripple in the characteristics for v versus ϕ̇ and i versus ϕ due to the finite number of coils
in a dc machine, but (10.10) are accurate enough for many design purposes.
where KE is the back-emf constant of the dc machine and KT is its torque constant . The designation of
the former constant comes from the custom to call the voltage v induced in the dc machine rotor back
electromotive force (back emf). It is important to note that both constants are exactly the same.
Their
values are only equal when self-consistent unit system is used. Expressed in the SI units, KE V.s.rad−1
and KT N.m.A−1 . Thus both the constants can be replaced in (10.7) by a constant KEM called
electromechanical constant . This constant is determined by the machine dimensions and by the state of
its magnetisation.
Fig. 10.5a shows a configuration of a permanent magnet dc machine in which Ra and La represents
the armature circuit resistance and inductance, respectively. Naturally, we can add these elements also
to the basic ideal-transformer model of the machine together with the elements modeling mechanical
friction br of the rotor bearing and the rotor inertia J as shown in Fig. 10.4b.
10-4 MODULE 10. ELECTROMECHANICAL SYSTEMS
We shall replace the condenser by a pure capacitor assuming that the phenomena like electrical
inductance and resistance or mechanical mass and friction could be modeled by additional physical
elements.
The constitutive relations for the movable-plate capacitor can be expressed as
In electrically linear case, the first of the relations (10.12) has the form v = q/C(x), where C(x) is the
capacitance when the movable plate is in the position x. In general, C(x) is a nonlinear function of x.
An important fact is that when q = 0, the force F must be zero for any x. If there is no charge, there is
no electric field, and hence no attractive force between the plates.
The electrical and mechanical power delivered to the capacitor in time dt represents the increase dWe
in its stored electrical energy, i.e.
∂We ∂We
dWe = vi dt + F ẋ dt = v dq + F dx = dq + dx (10.13)
∂q ∂x
Because of its conservative nature, the total electrical energy of the capacitor We (x, q) is obtained by
integrating (10.13) along any path from the datum configuration with x = 0 and q = 0. If we integrate
(10.13) from (x = 0, q = 0) to a configuration (q, x) along the rectangular path following first the x-axis
up to the point (q = 0, x), and then the q-axis up to (q, x), the electrical energy is, simply,
Z q
q dq q2
We (q, x) = = (10.14)
0 C(x) 2C(x)
10.6. MOVABLE-CORE SOLENOID 10-5
as there is no contribution from the segment along the x axis (where the force F is zero for any x).
Once the energy function We (q, x) is known, the constitutive relations (10.12) can be recovered by its
differentiation:
∂We q ∂We q2 C ′
v= = F = =− (10.15)
∂q C ∂x 2C 2
where the derivative C ′ (x) = dC(x)/dx.
Figure 10.7: (a) Energy-storing transformer, (b) its source implementation. (c) Energy-storing gyrator,
(b) its source implementation.
dq 1
i= = C(x) v̇ + K(x, v) ẋ F = − K(x, v) v (10.16)
dt 2
where K(x, v) = C ′ (x) v. Than it should be seen that a variable pure capacitor combined with a pure
energy-storing gyrator can model the electro-mechanical coupling as shown in Fig. 10.6b. Fig. 10.7d
shows that the energy-storing gyrator consistscan be formed from a controlled source of current, and of
a controlled source of force.
Replacing the coil by a pure inductor, we can express its constitutive relations in the form of equations
for its current i and the external force F (required to hold the core in equilibrium against the magnetic
attraction) in terms of the inductor flux linkage λ, and the movable core displacement x as
If the relation between current and flux linkage is linear, the first equation becomes i = λ/L(x) where
L(x) is the inductance of the coil when the core is in the position x. When λ = 0, there is no magnetic
field, and hence no magnetic attraction. The force F is therefore zero for any x when λ = 0.
Using a procedure analogous to that used in the previous example, the total magnetic energy of
the inductor Wm (λ, x) can be obtained by integrating the increase dWm in its stored magnetic energy
corresponding to the electrical and mechanical power delivered to the inductor in time dt along a path
from the datum configuration (λ = 0, x = 0). In this way we find
λ
λ dλ λ2
Z
Wm (λ, x) = = (10.18)
0 L(x) 2L(x)
Again, the constitutive relations (10.17) can be recovered by differenting the energy function Wm (λ, x):
∂Wm λ ∂Wm λ2 L′
i= = F = =− 2 (10.19)
∂λ L ∂x 2L
di 1
v = λ̇ = L(x) + K(x, i) ẋ F = − K(x, i) i (10.20)
dt 2
where
K(x, i) = L′ (x) i (10.21)
Hence, the corresponding pure model can be of the form shown in Fig. 10.8b. It consists of a variable pure
inductor combined with an energy-storing transformer the implementation of which in terms of controlled
sources is shown in Fig. 10.7b.
N 2 µ0 A N 2 µ0 A
L(x) = , the factor K(x, i) = − i
x x2
where N is the number of winding turns, x is the air-gap width, A is the cross-sectional area of the
air-gap, and µ0 is the air permeability. The electrical resistance of the relay winding, the mass of the
movable part, its friction, and the spring effect are represented in the model by physical elements.
Fig. 10.9c shows a relay model in which the magnetic circuit is represented in a more authentic way.
Coupling between the electrical and magnetic circuits is modeled there by a gyrator of gyration ratio
equal to N . The elements Pf and Pa represent permeances of the ferromagnetic part of the magnetic
circuit and of the air-gap, respectively. The magneto-mechanical coupling is modeled by an energy-storing
gyrator. The electrical voltage is replaced there by magnetic voltage, the electrical charge by magnetic
flux, and the electrical capacitance by permeance. In this case,
wa
K(x, wa ) = P ′ (x) wa =
µ0 A
Figure 10.9: (a) Electromagnetic relay, (b) electro-mechanical and (c) electro-magneto-mechanical mul-
tipole model.
In general, for an n-phase motor, the stator teeth are displaced by 1/n of a tooth pitch from section to
section.
To form a multipole model of the stepping motor, we need again to know the form of its inductance.
In practice, the motor inductance L as a function of displacement θ may be again approximated by a
cosine wave, that is,
L(θ) = L1 + L2 cos rθ (10.23)
where L1 and L2 are constants and r is the number of teeth on each rotor section.
Now, in the case of a three-phase motor with an 10-teeth rotor, r = 10, let the equilibrium position
be the situation when phase A is energized. Then the inductance for phase A is given by
Phase B has its equilibrium position 120◦ behind the reference point. Therefore, the inductances and
the K-factors for phases B and C are then written as follows:
The electrical circuits of the three phases are isolated so that each phase can be modeled in a separate
way. The total torque developed on the rotor is the algebraic sum of torques of the three phases. Thus
τ = τA + τB + τC (10.25)
10-10 MODULE 10. ELECTROMECHANICAL SYSTEMS
Module 11
1. the energy interactions between subsystems take place just in a limited number of interaction sites
formed by adjacent energy entries into the subsystems
2. the energy flow through each energy entry can be expressed by a product of two complementary
quantities called power variables
3. the total energy flow into a subsystem equals to the sum of the energy flows through all its energy
entries
The energy interactions in our examples are assumed to take place at the sites denoted by M, N
and and 0. In the case of the electrical system, these sites are made by interconnections of the energy
entries formed by electrical terminals of the system parts. In the fluid system, the energy entries are
formed by pipe inlets of the parts. The spring and the damper in the translational system interact via
common cross-sections of their links. We can assume that the cable, the spring and the body interact via
a cross-section of the body. We could easily identify similar interaction sites and energy entries of parts
in the rotational system
11-1
11-2 MODULE 11. UNIFIED APPROACH TO MODELING
Table 11.1: Examples of simple dynamic systems and their multipole models.
Configurations of real systems Multipole diagrams
M part3 N
M 3 L
1 part1 part2 part4
2 C G
4 N J
0
M N
2
3 N part3
part1 L part4
1
M J part2 G
0 4 C
M N
part3 part4
part1
L G
J part2
C
part3
M part1 part4
1 N part2 L
3 J G
0 j 4 C
2
Part Real parts Physical elements
1 combustion engine source of torque
2 flywheel rotational inertor
3 long flexible shaft torsional spring
4 ventilator propeller rotational damper
11.1. DYNAMIC BEHAVIOR OF SYSTEMS 11-3
related pairs of energy variables, i.e. the through and across variables integrated with respect to time.
The vacancies in Table 11.2 correspond to quantities which have no physical meaning.
Note that the through and across variables differ by the way in which they are directly measured.
Through variables are measured between interacting adjacent energy entries by instruments included
between the entries after the entries were disconnected. Across variables are measured between dis-
tant entries by instruments applied to the entries without any disconnections (across variables between
adjacent entries are zero). This is illustrated by Fig. 11.6 showing a two-entry part connected to a system.
An important role is played by across variables measured with respect to an absolute across-variable
reference. In each of the examples given in Table 11.1, such a reference is denoted by 0. The absolute
reference of voltages in electrical systems – electrical ground – is embodied in the example by the system
chassis. In the fluid system, the free atmospheric pressure was chosen as the pressure reference. In
both the mechanical systems, the system frame fixed to the Earth surface is considered as the absolute
reference for velocities.
The third approximating assumption mentioned above can be expressed for a subsystem with n entries
as
n
X
P (t) = ij (t) · vj (t) (11.1)
j=1
where P is the total power consumed in the subsystem, ij is the through variable of the j-th entry, and
vj is the across variable of this entry with respect to the absolute reference, j = 1, 2, . . . , n. We thus
A B
iB
PART A PART B
A PART B'
vBA B vA vB
(a) (b) (c)
Figure 11.1: Measuring of (a) an across variable between distant energy entries, (b) a through variable
between adjacent entries, (c) across variables of entries with respect to a reference.
11-4 MODULE 11. UNIFIED APPROACH TO MODELING
assume that the power consumed (dissipated or accumulated) in a subsystem is positive whereas the
power retrieved from the subsystem is negative.
• Orientation of across variables in a system is indicated by arrows with empty arrowheads. These
arrows are always directed from the point of the assumed smaller value to the point of the assumed
larger value of the related across variable with respect to its absolute reference.
• Orientation of through variables is indicated by arrows with full arrowheads. The full-head arrow
direction is in agreement with the assumed positive direction of the motion of the associated medium.
Thus the empty-head arrow points in the direction of the assumed increasing value of the across
variable. The full-head arrows point in the direction of the assumed positive motion of a flow of positively
charged particles in the electrical domain, magnetic flux in the magnetic domain, heat flow in the thermal
domain, and flow of mass particles in the fluid or acoustic domain. In the mechanical domain, a full-head
arrow indicates the direction of the motion the related force or torque tends to cause.
The links can be viewed as interconnections capable of transferring energy in both directions without
any dissipation, accumulation or delay. Depending on the related physical domain, the links can be
viewed either as idealized electricity-conducting wires or as fluid-conducting pipes, idealized shafts or
other mechanical links, etc. This does not mean, however, that real interconnections should be always
modelled in such an idealized way; they can be considered as independent subsystems with any of their
non-ideal features respected. This is the case of the long pipe segment in the fluid-system example, and
of the long torsionally flexible shaft in the rotational-system example in Table 11.1.
11.2. DISSIPATIVE AND ACCUMULATIVE ELEMENTS 11-5
Each of the physical elements is characterized by a specific constitutive relation between the el-
ement variables. Constitutive relations for the dissipative and accumulative elements are given in
11-6 MODULE 11. UNIFIED APPROACH TO MODELING
electrical
magnetic
thermal
fluid or
acoustic
translational
rotational
generic
Table 11.5. The letter p denotes the element parameter p in these relations. Table 11.5 gives also
physical dimensions of the parameter p in different physical domains and their units. The fact that the
constitutive relation of the dissipative and accumulative elements are in the form of algebraic of ordinary
differential equations indicates that the elements belong to lumped-parameter models.
In the case of ideal elements, the parameter p is constant. The elements given in Table 11.4 need
not be ideal, however. DYNAST allows also for nonlinear and time-variable elements as well as for
elements the parameters of which are are controlled by variables or parameters of some other elements,
blocks or equations.
In general, the parameter p of a controlled nonlinear and time-variable element can be given by an
expression of the form
p = f (z1 , z2 , . . . , ż1 , ż2 , . . . , t) (11.3)
where z1 , z2 , . . . are elementcontrolling variables or parameters, ż1 , ż2 , . . . are their time derivatives, and
t denotes time.
Despite the fact that all the physical elements are two-pole models, one-pin symbols are given in
Table 11.4 for thermal, fluid or acoustic capacitors as well as for the translational and rotational inertors.
According to physical laws governing accumulation of energy by these elements, their element-across
variable ve must be always considered with respect to the related reference. In other words, one pole of
11.3. SOURCES OF ENERGY 11-7
d d
Constitutive relation ie = p ve ve = p ie ie = p dt ve ve = p dt ie
these elements must be always coalesced with a reference node. Only one pin needs to be shown in the
graphical symbols of these elements to simplify the multipole diagrams.
Mechanical
source of force source of velocity
translational
rotational
Constitutive relation p = ie p = ve
Figure 11.3: (a) Interaction between a system and its surroundings modelled by (b) across-variable source,
or (c) through-variable source.
• An independent source of magnetic voltage can be used to model generation of magnetic field in a
permanent magnet.
• In the analysis of the heat dissipation from a system in a room, the air in the room can be treated
as an independent source of temperature if the heat transfer from the system to the room air is not
large enough to have a noticeable effect on the room air temperature.
• Most of the hydraulic pumps are usually modelled as independent sources of volume flow.
• A large accumulator supplying fluid to a small fluid-power system may be replaced by an indepen-
dent source of pressure.
• A sufficiently powerful motor driving a light mechanism represents from the point of view of the
mechanism an independent source of angular velocity.
• When modeling motion of a body close to the earth surface, the effect of the earth gravitational
field on the body can be respected by an independent source of gravitational force Fg = g · m, where
g is the gravitational constant and m is the mass of the body.
It is assumed in each of the examples given in Table. 11.1 that the system under our investigation
consists of parts 2, 3 and 4 while the part 1 is considered there as the system surroundings. For this
reason, each of the parts 1 is modelled by an independent source of through variable. If modeling of a
real source or the system surroundings by an independent source alone is inadequate, the independent
source is combined with other physical elements.
11.3. SOURCES OF ENERGY 11-9
Figure 11.4: Characteristic of ideal (a) closed and (b) open connection.
Note also, that the zero-valued independent sources can be used to imitate measuring instruments
and sensors of power variables. An ideal through-variable indicator models ideal through-variable
measuring or sensing devices. It behaves like an ideal connection the through-variable of which corre-
sponds directly to the indicated quantity. Similarly, an ideal across-variable indicator acts as an ideal
open connection and indicates its own across variable. Therefore, the ideal through-variable indicator is
equivalent to an independent source of zero across variable, whereas the ideal across-variable indicator is
equivalent to an independent source of zero through variable.
Figure 11.5: Graphical symbols for an ideal (a) across-variable and (b) through-variable indicator.
In cases when we shall need to stress the presence of ideal measuring instruments in a system model,
we shall use the graphical symbols shown in Fig. 11.5. Each of the symbols for the ideal indicators
represents a two-pole with its +pole at the top.
Fig. 11.6 shows configurations in which ideal indicators are used to ‘pick-up’ power variables of a
generic physical element in a system. The instrument orientation with respect to the element orientation
was chosen in such a way that the variable polarities are in mutual agreement.
Figure 11.6: Configurations for indicating (a) element-across variable and (b) element-through variable.
11-10 MODULE 11. UNIFIED APPROACH TO MODELING
In general, the controlled sources of through variables are characterized by the constitutive relation
The counterparts, controlled sources of across variables are characterized by the relation
• The element-across variable ve is the across variable of the element +pole with respect to its −pole
(or to the absolute reference if the second pin is not shown in the element symbol).
• The positive orientation of the element-through variable ie is always directed from the +pole towards
the −pole.
Fig. 11.2 shows that according to our conventions the empty-head arrow for ve is always directed
from the −pole towards the +pole of the element, whereas the full-head arrow for ie is always directed in
the opposite way. The empty-head arrow for the pole variable v+ and v− is always directed away from
the absolute reference towards of the +pole and −pole, respectively. Note, however, that once we have
interrelated the variable orientation with the element polarity, we do not need to use the arrows at all.
to the across variable vBA of the entry B with respect to the entry A as indicated in Fig. 11.7a. Similarly,
the orientation of the element polarity with respect to the entries A and B determines orientation of the
element-through variable ie . It corresponds in this case to the through variable entering the subsystem
at the entry A and leaving it at the entry B.
If we had chosen the opposite orientation of the element with respect to the subsystem entries, also
the orientation of the variables ie and ve would be opposite. Note that the presumption (11.2) is satisfied
regardless of the chosen element orientation with respect to the entries.
Figure 11.7: (a) Subsystem with two energy entries of non-mechanical nature, (b) its generic-element
model.
Figure 11.8: (a) Subsystem with two energy entries in translational motion, (b) its generic-element model.
Fig. 11.9 shows the similar situation for a subsystem with entries in rotational motion. In summary,
the mutual orientation of variables associated with a mechanical subsystem and its elemental model
must be related to the mutual orientation of the subsystem coordinate axes and the element polarity.
Whenever possible, we shall always prefer the mutual subsystem-element orientation shown in Fig. 11.8b
and Fig. 11.9b.
(a) (b)
Figure 11.9: (a) Subsystem with two entries in rotational motion, (b) its generic physical-element model.
Figure 11.10: (a) Force actuator, (b) and (c) corresponding sources of force.
Source of velocity imposes a relative motion of a specified velocity between two points it represents.
Fig. 11.11a shows point A that is in a rectilinear motion along the x-axis of a given absolute velocity
ẋA (t). In Fig. 11.11b the x motion of the point A is represented by node Ax , the actuator giving the
velocity ẋA (t) to the point A an ideal source of velocity modeling motion of a point A along the x-axis
with known absolute velocity ẋA (t).
The source of velocity shown in Fig. 11.11b models imposing a relative motion of point B with respect
to point A with known relative velocity ẋBA (t) along the x-axis as shown in Fig. 11.11a.
(a) (b)
Figure 11.11: (a) Moving points, (b) source of velocity modeling an imposed motion of the points.
• Define the system and its surroundings and denote by identifiers the sites of energy interaction
between them
• Decompose the system into subsystems and denote the sites of energy interaction between the
subsystems
• Associate all the sites of mechanical-energy interaction with a suitable coordinate system
• Start sketching the multipole model of your system by placement of nodes representing the indi-
vidual sites of energy interaction
• Choose a suitable physical-element model for each subsystem and place it between the nodes cor-
responding to the subsystem interaction sites
11.5. CONFIGURATIONS OF PHYSICAL ELEMENTS 11-13
• Orient the polarity of each mechanical element in conformity with the chosen coordinate system
• You may add ideal indicators for the across and through variables you are interested in
Though the symbols were chosen to remind real subsystems you should never forget about the dif-
ference between a real subsystem and its model. As a matter of fact, no real subsystems behave exactly
like the physical elements. The elements are, however, very useful first-order approximations of dy-
namic effects taking place in real subsystems. If modeling of a subsystem by a single element appears to
be inadequate, a more realistic subsystem model can be constructed using several physical elements as
standardized building bricks.
for any t. Otherwise the variable would be inconsistent. If, however, (11.6) is satisfied, any one of
the three across-variable sources can be omitted without any impact on the system-model dynamics.
Therefore, if (11.6) is not satisfied, the equations characterizing the system configuration in Fig. 11.12a
have no solution. If (11.6) is satisfied, the equations are singular unless the superfluous source is removed
from the system model.
Figure 11.12: Forbidden (a) loops of across-variable sources and (b) cut-sets of through-variable sources.
The three through-variable sources in Fig. 11.12b form a cut-set, or – in other words – if the three
sources are removed from the system model, the model gets separated into two isolated parts. At the
same time, according to Postulate of Continuity the element-through variables of the sources must for
any t satisfy the equation
i1 (t) − i2 (t) + i3 (t) = 0 (11.7)
Then again, if (11.7) is satisfied any one of the sources can be omitted in the configuration without
affecting the system-model behavior. Also in this case, unless the superfluous source is removed from the
system model, the equations characterizing the system are either having no solution or they are singular.
This observation, that can be easily proven in a general way, results in the following conclusion:
Loops of across-variable sources and cut-sets of through-variable sources are forbidden in the system-
model configurations.
Fig. 11.13 and Fig. 11.14 give examples of permitted and forbidden source configurations. Explain
the reason why the configurations in Fig. 11.14 are forbidden and try to submit them to DYNAST to see
what kind error messages DYNAST will give you.
11-14 MODULE 11. UNIFIED APPROACH TO MODELING
Let us consider the four-pole shown in Fig. 11.15 which consists of two generic two-pole elements
which are mutually controlled by each other’s element variables. Assuming that the coupling is pure, i.e.
11.6. ENERGY TRANSDUCERS 11-15
that the total power lost or accumulated in the coupling is zero, the across and through variables va (t)
and ia (t) of the element a together with the analogous variables vb (t) and ib (t) of the element b must
justify the energy-transfer relation
With regards to signs of the variables, we have respected here the Element-Variable Polarity Convention.
where f (.) is a time-independent function and qa (t), qb (t) are time-integrated element-through variables
of the elements a and b, respectively (in mechanics such couplings are called holonomous scleronomous).
The variables come possibly but not necessarily from different energy domains.
If the function f (.) is differentiable, we can apply the chain rule to differentiate (11.9) with respect
to time to obtain
∂f dqa ∂f dqb
· + · =0 (11.10)
∂qa dt ∂qb dt
Taking into consideration that dqa /dt = ia (t) and dqb /dt = ib (t), (11.10) can be rewritten as
where
na
n= [−] (11.13)
nb
is the dimensionless coupling ratio of the energy-transfer transformer.
To arrive at the same pure energy converting model, Instead from (11.9) we could have started the
derivation of the energy-transfer transformer constitutive relation from the describing relation
g λa (t), λb (t) = 0 (11.14)
specifying a coupling between the time-integrated element-across variables λa (t), λb (t) of the elements a
and b. Following a similar procedure, in this case we would arrive at the constitutive realtion
ia (t) 0 m va (t)
= · (11.15)
vb (t) −m 0 ib (t)
where
nb 1
m=− =− (11.16)
na n
Obviously, m is a negative reciprocal of the coupling ratio n defined by (11.13).
Fig. 11.16 presents examples of pure energy-transfer transformer implementations in which the generic
elements a and b are replaced by controlled sources. The source combinations shown in Fig. 11.16a and
b correspond directly to (11.12) and (11.15). The source combinations given in Fig. 11.16c and d must
be considered simultaneously with (11.11) and 11.8). Thanks to the implicit form of the equations, the
latter transformer models are more flexible from the point of view of their interconnection with other
multipoles in a system models.
Graphical symbols which we shall use for pure energy-transfer transformers are indicated in Fig. 11.17.
In the case of the symbol with the slash sign shown in Fig. 11.17a the parameter k equals to the coupling
ratio (11.13), i.e., k = na /nb , whereas in the case of the symbol with the backslash sign shown in
11-16 MODULE 11. UNIFIED APPROACH TO MODELING
Figure 11.17: Graphical symbol for energy-transfer transformers with (a) k = na /nb and (b) k = nb /na .
Fig. 11.17b the parameter k is defined by the reciprocal expression, i.e., k = nb /na . We are assuming,
that the index a is always associated with the left-hand pair of poles of the symbol, whereas the index
b always belongs to the right-hand pole-pair. For this reason, the symbol must be always situated
horizontally in a multipole diagram.
An ideal energy-transfer transformer is the pure transformer the coupling ratio of which is a con-
stant. We shall encounter numerous examples, however, in which the ratios are functions of some system
variables or parameters.
Figure 11.18: (a) Cascade interconnection of energy-transfer transformers, (b) one-transformer equivalent.
Fig. 11.18a shows a ‘cascade’ interconnection of energy-transfer transformers of coupling ratios n1 and
n2 . A simple analysis of the total transformation of across and through variables in the interconnection
indicates that the cascade can be replaced by only one transformer with the coupling ratio n1 n2 shown
in Fig. 11.18b.
Fig. 11.19 indicates the way in which series or parallel combinations of physical elements connected
to a pole-pair of an energy-transfer transformer are ‘seen through’ the transformer, or referred to at the
other transformer pole-pair. Z resp. Y is an impedance resp. admittance of the generic element, E resp.
J is an independent source of an across resp. through variable. This is the principle behind utilizing
various transducers and couplings to convert the amount of an actuator load at the price of changed
related across and through variables.
11.6. ENERGY TRANSDUCERS 11-17
Fig. 11.20 presents examples of pure energy-transfer transformer implementations in which the generic
elements a and b are replaced by controlled sources. The source combination shown in Fig. 11.20a
corresponds directly to (11.20), Fig. 11.20b corresponds to the inversion of (11.20).
11-18 MODULE 11. UNIFIED APPROACH TO MODELING
Graphical symbols which we shall use for pure energy-transfer transformers are indicated in Fig. 11.21.
An ideal energy-transfer gyrator is a pure gyrator the gyrating ratio of which is a constant. We shall
encounter numerous examples, however, in which the ratios are functions of some system variables or
parameters.
Figure 11.22: (a) Cascade interconnection of an energy-transfer gyrator and transformer, (b) gyrator
equivalent.
Fig. 11.22a shows a ‘cascade’ interconnection of an energy-transfer gyrator of gyrating ratio S and
an energy-transfer transformer of coupling ratio n. A simple analysis of the total transfer of across and
through variables through the interconnection indicates that the cascade can be replaced by only one
gyrator with the gyrating ratio Sn shown in Fig. 11.22b.
Fig. 11.23 indicates the way in which series or parallel combinations of physical elements connected to
a pole-pair of an energy-transfer gyrator are ‘seen through’ the gyrator, or referred to at the other gyrator
pole-pair. Note that gyrators convert inductors or springs into capacitors or inertors and capacitors or
inertors into inductors or springs. This principle can be utilized both in system analysis as well as in
system construction.
11.6. ENERGY TRANSDUCERS 11-19
assuming that ik > 0 if ik leaves the node, and ik < 0 if ik enters the node.
(c) Postulate of compatibility resulting from the laws respecting the geometric connectedness of real
systems given in Table ??. According to this postulate the sum of all the elemental across variables vk
along a closed oriented loop in the diagram
X
vk = 0 (12.2)
k
where vk > 0 if the vk direction is in agreement with the loop orientation, and vk < 0 if the vk direction
is opposite to the orientation of the loop.
To demonstrate the principles of such equation formulation, let us start with the examples presented
in Table 11.1 already.
(a) Constitutive relations of physical elements involved in the examples are given in Table 12.1 in
three different forms where J, C, L and G are parameters of the elements, ik is the through variable, vk
is the across variable of the k-th element in each of the diagrams. Let us choose the implicit form:
i1 − J = 0
d
i2 − C dt v2 = 0
(12.3)
d
v3 − L dt i3 = 0
i4 − G · v4 = 0
12-1
12-2 MODULE 12. FORMULATION OF SYSTEM EQUATIONS
(b) Fig. 12.1 shows the graph of the multipole diagrams in Table 11.1. The graph nodes are identical
with the nodes of the diagrams. Each of the diagram elements is represented in the graph just by
a line segment. In Fig. 12.1(a) the line segments are associated with arrows indicating orientation of
the elemental through variables (fixed to the orientation of the individual elements). According to the
postulate of continuity (12.1) for the elemental through variables at nodes M and N
node M: i1 + i2 − i3 = 0
(12.4)
node N: i3 + i4 = 0
(c) Fig. 12.1(b) shows two independent loops in the graph. The loop a is made up by elements 1 and
2, and the loop of elements 2, 3 and 4. Arrows indicate the chosen orientation of the loops with respect
to the orientation of the elemental across variables. According to the postulate of compatibility (12.2),
the elemental across variables along the loops a and b
loop a: v1 − v2 = 0
(12.5)
loop b: v2 + v3 − v4 = 0
i3 v3
M N M N
i2 i4 b
v2 v4
i1 v1 a
0 (a) 0 (b)
Figure 12.1: Graph of the multipole diagrams in Table 11.1 with arrows indicating orientation of elemental
variables and chosen loops.
The four constitutive relations (12.3) together with the two continuity relations (12.4) and two com-
patibility relations (12.5) form eight equations for eight unknown variables. In this case, the number of
equations can be reduced by one after substituting i1 = J to (12.4).
In general, for a model composed from m physical elements, the described procedure, called also the
tableau formulation, results in 2m differential equations to be solved to obtain 2m unknown system
variables.
Examples
These steps can be illustrated again with the system models shown in Table 11.1.
Step 1. This step results in equations stated already in (12.4).
Step 2. Substituting the explicit constitutive relations for the elemental through variables from Table 12.1
into (12.4) gives
d
v2 − L1 v3 dt − i3 (0) = 0
R
J + C dt
(12.6)
1
R
L v3 dt + i3 (0) + Gv4 = 0
Step 3. The elemental across variables expressed in terms of the nodal across variables vM and vN are
v1 = v2 = vM
v3 = vN − vM (12.7)
v4 = vN
d
vM − L1 (vN − vM )dt
R
C dt = i3 (0) − J
(12.8)
1
R
L (vN − vM )dt + GvN = −i3 (0)
where the known variables were shifted to the right-hand sides of the equations.
Step 4. Using the nodal formulation, we have arrived at only two equations (12.8) to be solved simul-
taneously for the two unknown nodal across variables vM and vN .
Step 5. The elemental across variables are expressed in terms of the nodal across variables as in (12.7).
Step 6. The elemental through variables can be evaluated using the explicit constitutive relations given
in Table 12.1.
Note, however, that the nodal formulation cannot be applied directly to models with physical ele-
ments like sources of across variable, for example, as the explicit constitutive relation for their through
variables necessary in Step 2 of the formulation does not exist. Also, the integro-differential form of the
equations (12.8) does not suit to common computer computational routines for simultaneous solving of
the equations.
Step 1. For each of the ℓ loops write the compatibility equation (12.2).
Step 2. Express all the elemental across variables in the compatibility equations in terms of the elemental
through variables using the constitutive relations of the individual elements.
Step 3. Write the elemental through variables in terms of the loop through variables.
Step 4. Solve the resulting ℓ equations for the ℓ loop through variables.
Step 5. Find the elemental through variables in the model as a difference of two of the loop through
variables.
Step 6. Express the across variables of the elements using their constitutive relations.
12-4 MODULE 12. FORMULATION OF SYSTEM EQUATIONS
Example
To illustrate these steps let us apply them to the system models shown in Table 11.1 in which the source
of through variable was replaced by the source of across variable characterized by the constitutive relation
v1 = E (12.9)
Step 1. The the compatibility equations for this example are already in (12.5).
Step 2. Substituting the explicit constitutive relations for the elemental through variables from Table 12.1
into (12.4) gives
1
R
E− C i2 dt − v2 (0) = 0
(12.10)
1 d 1
R
C i2 dt + v2 (0) + L dt i3 + G i4 = 0
Step 3. The elemental through variables expressed in terms of the loop through variables ia and ib are
i1 = −ia
i2 = ia − ib
(12.11)
i3 = −ib
i4 = ib
− C1
R
(ia − ib )dt = −E + v2 (0)
(12.12)
1 d 1
R
C (ia − ib )dt − L dt ib + G ib = −v2 (0)
where the known variables were shifted to the right-hand sides of the equations.
Step 4. Using the loop formulation, we have arrived at only two equations (12.12) to be solved simul-
taneously for the two unknown nodal across variables ia and ib . (By adding the second equation to the
first one the number of equations to be solved is reduced only to one in this special case.)
Step 5. The elemental through variables are expressed in terms of the loop through variables as in
(12.11).
Step 6. The elemental across variables can be evaluated using the explicit constitutive relations given
in Table 12.1 and (12.9).
Note, however, that the loop formulation cannot be applied directly to models with sources of through
variable as the constitutive relation explicitly expressing their through variables necessary in Step 2 of
the formulation does not exist. Also, the integro-differential form of the equations (12.8) does not suit to
common computer computational routines for simultaneous solving of the equations.
d
J + C dt v2 − i3 = 0
i3 + Gv4 = 0 (12.13)
d
v3 − L dt i3 = 0
When writing the elemental across variables in terms of the nodal across variables (12.7) we obtain
the three first-order algebro-differential equations for three unknown variables vM , vN and i3
d
C dt vM − i3 = −J
i3 + GvN = 0 (12.14)
d
vN − vM − L dt i3 = 0
d
where the symbol dt is replaced by the operator s assuming that the product s x should be interpreted
d
as dt x.
The left-hand-side square and right-hand-side column matrices of (12.15) can be arranged in the
tabular form convenient for manual formulation of equations
vM vN i3
M Cs −1 −J
(12.16)
N G 1
L −1 1 −Ls
The empty entries in the tables stand for zeros.
In a similar manner, the nodal formulation can be extended towards sources of across variables or any
other physical elements. In the next paragraphs, you will learn how to form the extended nodal equations
very easily following a straightforward pattern.
• sources of through variables, conductors, dampers, capacitors, and inertors if their through variable
is used to control an element in the model, or if it is requested as the output variable.
12-6 MODULE 12. FORMULATION OF SYSTEM EQUATIONS
a va ivb
Conductor va vb a 1
or damper a G −G
b −1
b −G G
b G −G −1
a va vb i
va vb
a 1
Resistor a 1/R −1/R
b −1
b −1/R 1/R
b 1 −1 −R
a va i vb
Capacitor va vb a 1
or inertor a Cs −Cs Cv(0)
b −Cs Cs −Cv(0) b −1
b Cs −Cs −1 Cv(0)
a va vb i
a 1
Inductor
or spring b −1
b 1 −1 −Ls −Li(0)
M i1 i2
Mutual L1 −M s −L1 i1 (0)
inductance −M i2 (0)
L L
−L2 i2 (0)
L2 −M s
−M i2 (0)
a i
Source of va vb
a 1
through a −J
variable b −1
b J
b 1 J
a va vb i
Source of
a 1
across
variable b −1
b 1 −1 E
12.3. EXTENDED NODAL FORMULATION 12-7
Source of through c ai vc vd i
vc vd
variable controlled by a 1
vc a g −g
across variable g b −1
b −g g
d b g −g −1
c ic a i ic i
Source of through ic
variable controlled by a 1
a β
through variable b −1
b b −β
d b β −1
c i a i va vb ic i
c
Source of across
variable controlled by a 1
through variable b −1
r
1 −1 −r
d b
c a i va vb vc vd i
Source of across
a 1
variable controlled by vc
across variable b −1
m
1 −1 −µ µ
d b
Ideal c va vb i
i
operational a 1
amplifier a b −1
d 1 −1
b
Step 2. If the number of he counted elements is m and the number of the non-reference nodes is n, draw
tables for the (m + n)-dimensional square and column matrices. Denote the first n columns of the square
table by the identifiers of the nodal across variables, and the remaining m columns by identifiers of the
computed through variables. At the same time, denote the first n rows of both tables by the identifiers of
the system model nodes and the remaining m rows by identifiers of the elements related to the computed
through variables.
Step 3. For each of the elements in the system model associate its poles a and b shown in Table 12.2
with the nodes of the system model. Add then the entries of the element submatrices to the related
entries of the resulting matrices. Ignore the entries related to the reference node.
For example, to form matrices (12.16) for the system models in Table 11.1, the poles of the involved
elements should be associated with the model nodes in the following way:
element J: a ↔ M, b↔0
element C: a ↔ M, b↔0
(12.17)
element L: a ↔ N, b↔M
element G: a ↔ N, b→0
Try now to form matrices (12.16) using the stamp submatrices of the elements.
va vb vc vd i
i a 1
c a b −1
Ideal
c −n
transformer
d n
d b 1 −1 −n n
va vb vc vd i
va vb vc vd
i a 1
a g −g
c a b −1
b −g g
Ideal gyrator c g −g
c g −g
d −g g
d −g g
d b g −g −1
Table 12.3 gives stamp matrices of controlled sources. The controlling through variables are denoted
as iC , while iC stands for the controlling across variables between system model nodes c and d.
a va
a 1
signal source
k va vb
b a a 1 −k
scalor
b va vb vc
a
c a 1 −1 1
comparator
b a va vb
a s −1 va (0)
integrator
b a va vb
a 1 −s −vb (0)
differentiator
b a vA vB x1 x2 x3
a 1 −b0 −b1 −b2 − sb3 −b3 x3 (0)
va b 3 s3 + b 2 s2 + b 1 s + b 0 1 −a0 −a1 −a2 − s −x3 (0)
= 3 s −1 x1 (0)
vb s + a2 s2 + a1 s + a0
s −a x2 (0)
transfer block
Table 12.1 shows the constitutive relations of the physical elements from our example in two different
explicit forms (except the through-variable source for which only one explicit form exists). So, we are
supposed to choose one of the explicit expressions for each of the elements and, at the same time, to
rearrange the relations (12.4) and (12.5) into the explicit form in such a way that there is just one relation
for each variable. Therefore, we are faced with the so called causality problem.
For computational reasons numerical integration is preferred to numerical differentiation in block-
diagram oriented packages. Let us just form the first three explicit relations as
i1 = J R
v2 = 1/C R i2 dt + v2 (t0 ) (12.18)
i3 = 1/L v3 dt + i3 (t0 )
Once we made this choice, it can be proven that the rest of the explicit relation must be formed in
the following way:
v4 = 1/G · i4
i2 = −i1 + i3
i4 = i3 (12.19)
v1 = v2
v3 = −v2 + v4
For example, if we had chosen the relation i4 = G v4 for the fourth element, we would not be able to
solve the problem. For some system configurations, even some of the integral relations must be replaced
by differential ones to solve the causality problem.
Step 5: Setting up the block diagram.
The resulting block diagram is shown in Fig. 12.2 using the DYNAST graphical notation. Each of the
blocks in the diagram represents just one of the explicit relations from (12.18) and (12.19) (no blocks are
needed for the trivial relations for i4 and v1 ).
12-10 MODULE 12. FORMULATION OF SYSTEM EQUATIONS
i1
i2 v1 = v2
v3 i4 = i3
J
1/C v4
1/L
1/G
• If we want to represent energy interactions of a real subsystem by a block, for most of the subsys-
tems we must have at our disposal two different blocks at least, each of them representing the same
constitutive relation, but expressing explicitly different variables. In the case of the multipole ap-
proach, only one model is sufficient for each subsystem, and this model is reusable in any consistent
system configuration.
• In the block-diagram model of a complete system, in addition to the blocks representing the individ-
ual subsystems, there must be also blocks respecting the postulates of continuity and compatibility.
The latter blocks are not reusable as they suit just to one specific system configuration. For the
multipole models the postulate-respecting relations can be formed automatically and no additional
blocks are needed.
• To choose which block should model a particular subsystem, and to form blocks respecting the
postulates, we have to solve the causality problem. In the multipole approach such a problem is
not encountered at all.
The manual procedure for setting up a block diagram for the physical-level model of a real-life system,
usually nonlinear, is much more involved than could be shown in this example, of course. However, there
is also a big difference between DYNAST and the common block-diagram oriented packages in the way
of solving equations.
Once a block diagram is submitted to DYNAST, a set of implicit algebro-differential equations is
formed, and these are solved simultaneously as it is the case of the equations for multipole models. The
common block-diagram oriented programs usually do not solve any equations in reality. They would not
treat the relations (12.18) and (12.19) as equations, but as assignments.
The variables are processed successively by one block only at a time at discrete time steps. The
process starts from blocks with no input like the block specifying i1 (t1 ) ⇐ J(t1 ) in Fig. 12.2. But this
is not so simple in the case of feedback loops in the block diagram. For example, the summing block
for the output variable i2 (t1 ) ⇐ −i1 (t1 ) + i3 (t0 ) cannot specify the correct value of this variable at the
end of the first time step, but only after a number of time steps and iterations. This problem is even
more difficult if there is no integrator delaying the variable propagation in the feedback loop by one step.
Solving such algebraic loop problems requires special techniques.
In many cases, the blocks in a block diagram must be rearranged before the simulation starts as their
order is critical to the validity of the computed results. There are, however, system models characterized
by systems of differential equations the order of which is time-variable. Solving such an change-of-order
problem requires reconfiguration of the block diagram not only prior, but also during the simulation.
As all the equations are solved simultaneously in DYNAST, there are no algebraic-loop problem. As,
in addition, the equations are solved in DYNAST directly in their natural implicit algebro-differential
form, there is no change-of-order problem. Not only this, the related computational procedures for such
equations are in principle more efficient and robust then those implemented in the common block-diagram
packages. Obviously, developing software for automated construction of physical-level system models as
a pre-processor for a common block-diagram oriented package is not the optimal way of coping with the
problem.
12.5. LAGRANGE’S EQUATIONS OF MULTIPOLES 12-11
fk (sj , t) = 0, k = 1, 2, . . . r (12.20)
Step 5. Differentiate (12.20) to obtain n − i relations for the subsystem geometric compatibility
d
fk (sj , t) = 0, k = 1, 2, . . . r (12.21)
dt
Step 6. Consider the ‘free’ diagram, i.e., consider the disconnected from a system. Show all the external
forces the system exerts on the to hold it in equilibrium.
Step 7. Formulate n Lagrange’s equations in the form
r
d ∂K ∂K ∂P ∂D X ∂fk
= + Qj − − − λk j = 1, 2, . . . , n (12.22)
dt ∂ ṡj ∂sj ∂sj ∂ ṡj ∂sj
k=1
where K, P, and D is the kinetic, potential, and dissipative energy of the. Qj is a generalized external
force associated with the generalized coordinate sj , whereas λk is the Lagrange’s multiplier representing
the generalized reaction force associated with the coordinate sk .
Step 8. By excluding the r multipliers λk from (12.22) obtain n − r equations for quasi-static balance
of inertial, potential and dissipative forces.
Figure 12.3: (a) Body with two revolute joints, (b) its multipole model.
The subsystem shown in Fig. 12.3a is a body constrained to a planar motion with two revolute joints
denoted by A and B. O denotes the mass center of the body. Each of the joints is represented by two
poles in the multipole model of the body given in Fig. 12.3b. Poles Ax and Bx represent translational
motions of the joints along the x-axis whereas the poles Ay and By along the y-axis. Pole Oϕ corresponds
to the rotational motion of the body about the mass center O.
12-12 MODULE 12. FORMULATION OF SYSTEM EQUATIONS
The body shown in Fig. 12.3a is of i = 3 degrees of freedom. Let us consider n = 7 coordinates:
xA , yA , xB , yB , xO , yO , and ϕ. Then the r = n − i = 4 geometric constraints can be expressed as:
f1 = xO − xB − lB cos ϕ = 0 f2 = yO − yB − lB sin ϕ = 0
f3 = xA − xO − lA cos ϕ = 0 f4 = yA − yO − lA sin ϕ = 0
By differentiating the constrains we obtain the first part of the constitutive relations:
ẋO − ẋB + lB ϕ̇ sin ϕ = 0 ẏO − ẏB − lB ϕ̇ cos ϕ = 0
(12.23)
ẋA − ẋO + lA ϕ̇ sin ϕ = 0 ẏA − ẏO − lA ϕ̇ cos ϕ = 0
The body kinetic and potential energies are
1 1 1
K= mẋ2O + mẏO
2
+ JO ϕ̇2 P = mgyO
2 2 2
where JO is the moment of inertia of the body with respect to its center of mass.
Substitution into (12.22) yields for the individual coordinates:
xA : 0 = FAx − λ3 yA : 0 = FAy − λ4
xB : 0 = FBx + λ1 yB : 0 = FBy + λ2
d d
xO : dt mẋO = −λ1 + λ3 yO : dt mẏO = −mg − λ2 + λ4
d
ϕ: J
dt O ϕ̇ = τO − λ l
1 B sin ϕ − λ l
3 A sin ϕ + λ2 lB cos ϕ + λ4 lA cos ϕ
Finally, after excluding the multipliers, we are arriving at the equations for the force dynamic equi-
librium forming the second part of the constitutive relations for the multipole model of the body with
two revolute joints
mẍO = FAx + FBx
mÿO = FAy + FBy − mg (12.24)
JO ϕ̈ = τO + FBx lB sin ϕ − FBy lB cos ϕ − FAx lA sin ϕ + FAy lA cos ϕ
Figure 12.4: (a) Slider-crank mechanism with a flexible connecting rod, (b) its multipole model.
Fig. 12.4a shows a slider-crank mechanism which is considered as massless, but its connecting rod is
assumed to be flexible with dissipative losses. In the corresponding multipole model given in Fig. 12.4b,
the rotational motion of the crank is represented by the pole Bϕ whereas the translational motion of the
slider represents the pole Ax .
The mechanism shown in Fig. 12.4a is of i = 2 degrees of freedom. Let us consider n = 4 coordinates:
xA , ϕA , ϕB and s. Then the r = n − i = 2 geometric constraints can be expressed as:
f1 = xA − lB cos ϕB + (lA + s) cos ϕA = 0
f2 = y0 + lB sin ϕB − (lA + s) sin ϕA = 0
By differentiating the constrains with respect to time we obtain:
ẋA + lB ϕ̇B sin ϕB + ṡ cos ϕA − (lA + s)ϕ̇A sin ϕA = 0
(12.25)
lB ϕ̇B cos ϕB − ṡ sin ϕA − (lA + s)ϕ̇A cos ϕA = 0
The potential and dissipative energies are in this case
1 2 1 2
P = ks D= bṡ
2 2
12.5. LAGRANGE’S EQUATIONS OF MULTIPOLES 12-13
where k and b is the stiffness and damping factor of the connecting rod, respectively.
Substitution into (12.22) yields for the individual coordinates:
xA : 0 = FAx + λ1
ϕB : 0 = τB + λ1 lB sin ϕB + λ2 lB cos ϕB
ϕA : 0 = −ks − bṡ + λ1 cos ϕA − λ2 sin ϕA
s: 0 = −λ1 (lA + s) sin ϕA − λ2 (lA + s) cos ϕA
Finally, after excluding the multipliers, we can complete the constitutive relations by the equations
13-1
13-2 MODULE 13. FORMULATION OF TRANSFER FUNCTIONS
is
a a
∆ = 11 12
= a11 a12 − a21 a22
a21 a22
Determinants of larger matrices can be computed by means of the Laplace expansion. In general,
expansion of the determinant ∆ of an n-dimensional square matrix A along its k-th row, resp. along its
l-th column, yields
n
X n
X
∆= akl (−1)k+l ∆kl , resp. ∆= akl (−1)k+l ∆kl (13.4)
l=1 k=1
where akl is the element in the k-th row and l-th column of the determinant, and ∆kl is the (n − 1)-st
order determinant obtained from the determinant ∆ by deleting the k-th row and l-th column of ∆.
For example, the determinant ∆ of the order n = 3 associated with the 3 × 3 matrix
a11 a12 a13
A = a21 a22 a23
a31 a32 a33
You may wish to show that expanding along another row or column gives the same result. This
method is valid for any n. Determinant evaluation using the Laplace expansion, however, is not suitable
for the analysis of large systems as it requires (n − 1)n! multiplications. While for n = 3 the number of
multiplications is 12, for n = 4 it becomes 72 already.
Fortunately, the system matrices are usually rather sparse so that the formulas (13.4) give us the
chance to take advantage of this matrix property. In general: the larger is the analyzed system, the higher
is the percentage of zero-value elements in its matrix A. By starting the determinant expansion from the
row or column with the least number of nonzero elements we can reduce the amount of computations
considerably.
∆i←j (s)
Hij = (13.6)
∆(s)
where ∆i←j (s) is the determinant of the matrix A(s) the i-th column of which has been replaced by the
j-th column of the v B(s).
y(s)
H(s) = (13.7)
F (s)
13.4. SOLVED EXAMPLES 13-3
by assuming that the force F (t) is the input and the displacement y(t) of the mass is the output.
The multipole model of the system is given in Fig. 13.1a. The node A represents the vertical motion of
the mass. The polarity of the input-force source has to comply with the chosen orientation of the velocity
ẏ(t). In this simple case only one equation is needed to describe the dynamics of the given system. This
is the equation in Fig. 13.1b are
ẏ F
A b + sm + k/s 1
The transfer-function H(s) can then be readily expressed as
1/s 1
H(s) = = 2 (13.8)
b + sm + k/s s m + sb + k
Mechanical rotational system. Consider the system shown in Fig. 13.2a. The system consists of a
load inertia and a viscous-friction damper driven by a rotational motor by means of a flexible shaft. The
motor is modelled as it is seen from the load.
System parameters:
τm = motor torque applied to the system
bm = motor damping (corresponding to the slope of its loading characteristics)
k = shaft torsional stiffness
∆21
H(s) = (13.10)
∆
where
b + k/s −k/s
∆ = m
∆21 = −| − k/s| (13.11)
−k/s bl + sJ + k/s
Thus
k
H(s) = (13.12)
s2 Jbm + s(kJ + bm bl ) + k(bm + bl )
L-R-C circuit. Consider the electrical circuit shown in Fig. 13.3a. The circuit consists of an inductor
L, a resistor R, and capacitor C.
If ei is assumed to be the input and v0 the output, then the transfer function of the system is assumed
to be
v0 (s)
H(s) = (13.13)
ei (s)
The multipole diagram of the system is given in Fig. 13.3b. It differs from the actual circuit diagram
by the voltage source representing the input excitation ei .
When formulating equations describing the system, we shall first take into consideration all the three
nodes shown in the diagram:
v1 v2 v3 i ei
1 1 −1
1
sL sL
2 −1 1
G+ −G
sL sL
3 −G G + sC
ei 1 1
1 −1
sL 0 1
sL
−1 1
G+ −G 0 = G 1 + sC + C/L
∆(s) = sL sL
sL
0 −G G + sC 0
1 0 0 0
1 −1
sL 1
sL
3+1 −1 1
G
∆31 (s) = (−1) 0 = sL
sL G+
sL
0 −G 0
13.4. SOLVED EXAMPLES 13-5
1
H(s) = (13.14)
LCs2 + RCs + 1
In this case we can considerably simplify the problem by formulating just two equations, one for the node
3 and the second one for the series combination of the elements ei − L − R. Then we shall obtain
v3 i ei
3 sC 1
ei − L − R 1 −sL − R 1
In this case
sC 1
∆(s) = = −(R + sL)sC − 1
1 −sL − R (13.15)
∆11 (s) = (−1)1+1 | − 1| = −1
The resulting transfer-function H(s) is identical to (13.14).
Seismograph. Fig. 13.4a shows a schema of seismograph. A seismograph indicates the displacement
of its case relative to the inertial reference. It consists of a mass m fixed to the seismograph case by a
spring of stiffness k, b is a linear damping between the mass and the case.
Let us define:
xi = displacement of the seismograph case relative to the inertial reference (the absolute frame)
x0 = displacement of the mass m within the seismograph relative to its case
y = x0 − xi = displacement of the mass m relative to the case.
Note that y is the variable we can actually measure.Since gravity produces a steady spring deflection,
we measure the displacement x0 of mass m from its static equilibrium position.
Considering xi as input an y as output, the transfer function is
k
y m
C M
M d
yM
vC m
k d
C
yC
(a) (b)
The multipole model of the seismograph is given in Fig. 13.4b. It is advantageous to take there as a
reference node the node C corresponding to the case instead of the node F corresponding to the actual
inertial reference. The transfer function can than be expressed as
vMC (s)
F (s) = (13.17)
vF C (s)
Exploiting the fact that the velocity between nodes F and C is constrained by the input velocity source
vi , we can write the matrix-form equations for the multipole diagram in the following way
13-6 MODULE 13. FORMULATION OF TRANSFER FUNCTIONS
vMC vF C vi
M b + sm + k/s −sm
F 0 1 −1
From this
sm s2
F (s) = − =− 2 (13.19)
b + sm + k/s s + sb/m + k/m
Note that
lim F (jω) = 0 and lim F (jω) = −1 (13.20)
ω→0 ω→∞
Thus for low frequency inputs xi the mass m follows the case up and down. However, if the input
variable has a sufficiently high frequency, then the mass m remains almost fixed with respect to the
inertial reference and the motion of the case indicates the
p relative motion between the case and the mass.
For this reason the undamped natural frequency ω0 = k/m is made small in seismographs.
Liquid-level system. Consider the system of two interacting tanks shown in Fig. 13.5a.
A1 A2
C1 = and C2 = (13.21)
ρg ρg
where A1 , A2 [m2 ] are horizontal areas of the tanks, ρ [kg.m−3 ] is the mass density of the liquid, and
−2
g [m.s ] is the acceleration due to gravity. The volume fluid flow into the first tank through the control
valve is Qi . The tanks are interconnected by a short pipe with a valve representing a restriction of fluid
R1 . The fluid resistance of the second tank outlet is R2 .
The task is to compute the transfer function between the intake flow Qi and the outlet flow Q2 which
can be expressed as
Q2 p2
F (s) = = (13.22)
Qi R2 Qi
where p2 is the liquid pressure at the bottom of the second tank.
Based on the assumption that the system is either linear (for the laminar flow in the restriction) or
linearized (for the turbulent flow), it can be modelled by the multipole diagram given in Fig. 13.5 (the
fluid conductances G1 = 1/R1 , G2 = 1/R2 ).
The port diagram equations are then
p1 p2 Qi
1 G1 + sC1 −G1 1
2 −G1 G1 + G2 + sC2 0
13.4. SOLVED EXAMPLES 13-7
From there
∆21 1
F (s) = G2 = 2
(13.24)
∆ R1 C1 R2 C2 s + (R1 C1 + R2 C2 + R2 C1 ) + 1
x k1 u1
k1 k2
m1 k2 d
x1 1 2
f1 F1 F1
u2
m2 m1 m2
x2
(a) (b)
Multivariable mechanical system. Consider the system shown in Fig. 13.6a. We assume that the
system, which is initially at rest, has two inputs F1 (t) and F2 (t) and two outputs x1 (t) and x2 (t). In this
problem we would like to demonstrate formulation of a multivariable system transfer-function description
x1 (s) F1 (s)
= H(s) · (13.25)
x2 (s) F2 (s)
The equations describing the multipole diagram model of the system given in Fig. 13.6b are
ẋ1 ẋ2 F1 F2
1 b1 + sm1 + k1 /s −b1 −1
2 −b1 b1 + sm2 + k2 /s −1
The transfer-function matrix H(s) can then be expressed as
" #
1/s det A11 (s) det A12 (s)
H(s) = · (13.27)
det A(s) det A21 (s) det A22 (s)
where
det A = ∆(s) = (b1 + sm1 + k1 /s)(b1 + sm2 + k2 /s) − b21 (13.28)
and
det A11 (s) = −(−1)1+1 ∆11 (s) = |b1 + sm2 + k2 /s| = −b1 − sm2 − k2 /s
det A12 (s) = −(−1)1+2 ∆12 (s) = −| − b1 | = b1
(13.29)
det A21 (s) = −(−1)2+1 ∆21 (s) = −| − b1 | = b1
det A22 (s) = −(−1)2+2 ∆22 (s) = −|b1 + sm1 + k1 /s| = −b1 − sm1 − k1 /s
Similarly, the equivalent moment of inertia and friction of the gear train referred to the load shaft
are
ωm
ωm M
M L
L
x x
(a) (b)
dm M L
L
M dl Fl
ωm
Jm
tau_m ml
x
(c) (d)
The system shown in Figs. 13.7a, 13.7b and 13.7c are all represented by a simple port diagram in
Fig. 13.7d. Elements τm , Jm and bm are modeling the drive motor there, and the elements Fℓ , mℓ and bℓ
model the mechanical load of the systems. The ideal transformer models the actual motion conversion.
13.4. SOLVED EXAMPLES 13-9
In case of the system shown in Fig. 13.7a, the transformer ratio n corresponds to the lead of the
screw, which is defined as the linear distance which the load travels per revolution of the screw. In the
other two systems n = r where r is the radius of either the pinion or of the pulley.
The equations describing the dynamics of the systems are following:
ωM vL F τm Fℓ
M b + sJm −n 1
L bℓ + smℓ 1 1
transformer n −1
The determinant of the system matrix is
∆ = bm + sJm + n2 (bℓ + smℓ ) (13.32)
Using this, we can easily derive the following transfer functions
vL n
=
τm bm + sJm + n2 (bℓ + smℓ )
ωM 1
=
τm bm + sJm + n2 (bℓ + smℓ )
F −n(bℓ + smℓ )
=
τm bm + sJm + n2 (bℓ + smℓ )
vℓ −n2
=
Fℓ bm + sJm + n2 (bℓ + smℓ )
From the second transfer function we can conclude that the equivalent torque τe , inertia Je and
damping be of the load that the motor sees is
τe = nFℓ , Je = n 2 m ℓ , and b e = n2 b ℓ
Thus any of the three rotary-to-linear conversion systems can be represented from the point of view
of loading the drive motor just by the elements τe , Je and be as shown in Fig. 13.7e.
Note also, that the third transfer function becomes
F 1
=−
τm n
for bm = Jm = 0 as could be expected.
From the last transfer function it is obvious, that the load sees the motor as shown in Fig. 13.7f where
the elements Fe , me and be are
Fe = τm /n, me = Jm /n2 , and be = bm /n2
Rotary-to-rotary motion conversion. Consider the system shown in Fig. 13.8a. In this system, a
load is driven by a motor through the gear train. Assuming that the stiffness of the shafts of the gear
train is infinite (there is neither backslash nor elastic deformation) and that the number of teeth on each
gear is proportional to the radius of the gear, obtain the equivalent moment of inertia and equivalent
friction referred to the motor shaft and referred to the load shaft.
In Fig. 13.8a, the numbers of teeth on gears 1, 2, 3 and 4 are N1 , N2 , N3 and N4 , respectively. The
angular displacements of shafts are ϕ1 , ϕ2 , ϕ3 and ϕ4 , respectively. Thus, ϕ2 /ϕ1 = −N1 /N2 = −n21 and
ϕ4 /ϕ3 = −N4 /N3 = −n43 . The moment of inertia and viscous friction of each gear train component are
denoted by J1 b1 , J2 b2 , J3 b3 , and J4 b4 , respectively. J1 b1 include the moment of inertia and friction of
the motor, J4 b4 include the moment of inertia and friction of the load.
The multipole model of the gear train system is given in Fig. 13.8b. The equations describing the
multipole diagram model of the system given in Fig. 13.8 are
ϕ̇1 ϕ̇2 ϕ̇3 τ2 τ4 τm τl
1 b1 + sJ1 n21 1
2 b2 + sJ2 1 n43
3 b3 + sJ3 1 1
T12 −n21 1
T34 −n43 1
13-10 MODULE 13. FORMULATION OF TRANSFER FUNCTIONS